You are on page 1of 77

Tallinn Technical University

Interoperability of
Geographical Information Systems

by

Hanno Kuus

THESIS

submitted in partial fulfillment of the requirements for the degree of


Master of Science in Engineering

Tallinn 2002
Declaration

I hereby declare that the current thesis ‘Interoperability of Geographical Information


Systems’ is my own original work, submitted to the Tallinn Technical University in
partial fulfillment of the requirements of the degree of Master of Science, and it has not
been submitted for reception of an academic degree before.

Hanno Kuus

Tallinn, May 2002


Interoperability of Geographical Information Systems

Hanno Kuus

Abstract

There are many Geographical Information System (GIS) applications in use today,
unfortunately we are also used to various problems regarding the spatial data. An
important aspect of every system is interoperability—the ability of GIS software to use
the spatial data, provided by other systems. It is also important not to overlook the
quality of the data, i.e. whether the accuracy, structure and content of the exchanged
set of data conform to our requirements.
The purpose of this thesis is to give an overview of two prominent methods for provid-
ing interoperability in Geographical Information Systems—data files and geospatial data
services. At first the reader is got acquainted with basic concepts of Geographical Inform-
ation Systems. In the next chapter the properties of several file formats are described
and compared for fitness for practical tasks.
Data files have been the traditional method for exchanging data among applications
for a long time but different services are getting more and more attention as spatial data
sources nowadays. This thesis focuses mainly to web services, developed by the Open
GIS Consortium (OGC). The Web Map Service (WMS) and Web Feature Server (WFS)
service standards are getting closer examination. A comparative view of two covered
interoperability methods (files and services) is also provided.
The following chapters concentrate to the implementations of the OGC Web Map
Service. Three different WMS servers are compared in regard of building various spatial
data exchange services. A detailed look to the architecture and functioning of one server
application is also provided. Another chapter is dedicated to WMS client applications,
where four programs are compared and analyzed for suitability for practical situations.
The thesis contains also descriptions of three projects that are using the Open GIS
Consortium Web Map Service specification for solving practical problems in the field of
GIS interoperability.
Andmevahetus geoinfosüsteemides

Hanno Kuus

Kokkuvõte

Geoinfosüsteemid on tänapäeval laialt levinud, kahjuks oleme harjunud ka mitme-


suguste probleemidega ruumiandmetega töötamisel. Üheks oluliseks küsimuseks ruumi-
andmete kasutamisel on andmevahetus—tuleb veenduda, et kasutatav tarkvara suudab
kasutada sisendandmeid teistest süsteemidest, pöörates seejuures tähelepanu ka andmete
kvaliteedile: kas andmete täpsus, sisu ja struktuur on sellised, nagu me ootame.
Käesoleva magistritöö eesmärgiks on anda ülevaade kahest geoinfosüsteemides kasu-
tatavast andmevahetuse viisist—andmefailidest ja andmevahetusteenustest. Töö algab
geoinfosüsteemide põhimõistete tutvustamisega, seejärel võrreldakse mõningate laialt
levinud failivormingute omadusi ning nende kasutatavust praktilistes ülesannetes.
Traditsioonlise failivahetuse kõrval on tänapäeval olulisteks ruumiandmete allika-
teks saamas ka andmevahetusteenused. Antud töös on tähelepanu pööratud peamiselt
OpenGIS konsortsiumi poolt loodud veebipõhiste teenuste tutvustamisele. Lähemalt on
kirjeldatud kaardiserveri (Web Map Service (WMS)) ja objektiserveri (Web Feature Ser-
ver (WFS)) teenuseid. Toodud on ka kahe käsitletud andmevahetusmeetodi (failide ja
teenuste) omavaheline võrdlus.
Magistritöö teises pooles on lähema vaatluse alla võetud kaardiserveri teenuse raken-
duslik külg. Võrreldud on kolme WMS serveri poolt pakutavaid võimalusi andmevahetus-
teenuse realiseerimiseks, detailsemalt on tutvustatud ühe serverprogrammi arhitektuuri
ning funktsioneerimist. Eraldi peatükk on pühendatud ka kaardiserveri klientprogrammi-
dele, kus on võrreldud nelja erineval tehnoloogial põhinevat rakendust ning analüüsitud
nende sobivust praktikas aset leidvatesse situatsioonidesse.
Magistritöö sisaldab ka kolme rakendusprojekti kirjeldust, kus on kasutatud Open-
GIS konsortsiumi kaardiserveri standardit reaalses elus esilekerkivate probleemide lahen-
damiseks.
Acknowledgments

No man is an island and this work could have not been produced without the assistance
of many people. First I would like to thank my supervisor, professor Vello Kukk, for his
encouragement and guidance throughout my studies. I am also thankful to professor Ennu
Rüstern, head of the Department of Computer Control, for his constant encouragement. My
appreciation goes to the whole academic staff of the Tallinn Technical University and the
Institute of Geography of the Tartu University for educating and directing me.
Special thanks belong to Peep Krusberg for numerous discussions and suggestions, and
to Peeter Väling for his help in finalizing the thesis. I would also like to thank my colleagues
in the R-Süsteemid company for many ideas and discussions, and for creating a pleasant
working atmosphere.
I am deeply grateful for my wife and son for their constant support and persuasion, and
for providing necessary relaxation.
Finally I should mention my thankfulness and respect to the numerous researchers who
have developed the huge body of knowledge that was available to me in printed form and
electronically in the Internet.
Contents

1 Introduction 1
1.1 Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2 Concepts of GIS 3
2.1 GIS Data Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Co-ordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Metadata . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3.1 Importance of Metadata: An Example . . . . . . . . . . . . . . . . . . 9
2.4 Standardization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3 Interoperability 13
3.1 File Exchange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.1.1 ESRI Shapefile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.1.2 Intergraph/Microstation Design File . . . . . . . . . . . . . . . . . . . 17
3.1.3 S-57 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.1.4 DIGEST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.1.5 GML . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1.6 Comparison of File Formats . . . . . . . . . . . . . . . . . . . . . . . . 23
3.2 Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2.1 Proprietary Map Servers . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.2.2 OpenGIS Web Services . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.2.3 OpenGIS Web Map Service . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.4 OpenGIS Web Feature Server . . . . . . . . . . . . . . . . . . . . . . . 31
3.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

4 WMS Servers 35
4.1 MapServer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.2 CubeSERV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.3 F-Open Server . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.3.1 History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.3.2 Frege . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.3.3 System Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.3.4 Future Plans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
CONTENTS vi

5 Web Map Service Client Programs 46


5.1 CubeVIEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
5.2 WMSClient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.3 WMS Client Toolkit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.4 FRED Map Editor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

6 WMS in Action 56
6.1 Public Map Service of the Estonian Maritime Administration . . . . . . . . . 56
6.2 Data Exchange between Map Authorities . . . . . . . . . . . . . . . . . . . . 57
6.3 The Registry of Assets of RMK . . . . . . . . . . . . . . . . . . . . . . . . . . 58
6.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

7 Conclusions 60

Glossary 62
Abbreviated Terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

Bibliography 65
List of Tables

2.1 Comparison of raster and vector models. . . . . . . . . . . . . . . . . . . . . . . . 5


2.2 Levels of data abstraction in GIS . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

3.1 Comparison of selected GIS file formats. . . . . . . . . . . . . . . . . . . . . . . . 23


3.2 Comparison of file and service based interoperability methods. . . . . . . . . . . 34

List of Figures

2.1 The components of a GIS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4


2.2 Co-ordinate systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3 Layered maps of the same area with different co-ordinate systems. . . . . . . . . 10

3.1 The overall concept of services. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26


3.2 Data flow in WMS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.3 A sample session between a WMS server and a client. . . . . . . . . . . . . . . . 29
3.4 A sample session between a WFS server and a client. . . . . . . . . . . . . . . . . 32

4.1 Handling maps with Frege. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39


4.2 The work-flow in the Frege presentation library. . . . . . . . . . . . . . . . . . . . 41
4.3 The schematic diagram of the F-Open server. . . . . . . . . . . . . . . . . . . . . 42

5.1 CubeVIEW screenshot. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47


5.2 WMSClient screenshot. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.3 Map browsing application of the registry of assets of RMK. . . . . . . . . . . . . 51
5.4 The result of the feature info request. . . . . . . . . . . . . . . . . . . . . . . . . 52
5.5 Screenshot of the FRED map editor. . . . . . . . . . . . . . . . . . . . . . . . . . 53
Chapter 1

Introduction

Today, computers are around us everywhere, performing tasks varying from simple to ex-
tremely complex. Geography is among the fields of research where computers are of great as-
sistance nowadays, especially since the advent of the Geographical Information System (GIS)
at the end of the 1950s, that revolutionized map-making, affected significantly the data ana-
lysis and visualization techniques, and, furthermore, opened many new opportunities to
make use of geospatial data. These days, in the era of personal computers, GIS is used in
many areas by a very wide range of specialists in their everyday work in solving multifarious
problems.
The commercial GIS industry, having its roots in the 1970s, is still growing at a steady
pace, about 20 per cent, each year [36]. This has led us to the situation where there are nu-
merous great software packages from various firms for building our Geographical Information
Systems. However, the standardization process in this area started only recently. The lack of
universally adopted standards has produced a situation, where the communication between
different systems is made hard and sometimes impossible. There have been efforts for creat-
ing standards but these activities have mostly remained local to some nationality or group of
interest. In the 1990s, at last, several international standardization projects took off. These
major players, like Open GIS Consortium (OGC), International Organization for Stand-
ardization (ISO) Technical Committee (TC) 211, Digital Geographic Information Working
Group (DGIWG) and others, have crafted several well-balanced standards for regulating the
GIS community. All these organizations are still busy in improving their specifications and
creating new standards, making this world a better place for all GIS users.
The author of this paper is working in the R-Süsteemid1 company, located in Tallinn,
Estonia. I have had a chance of working on several projects that used geographical data in
this small software engineering and consulting office. These projects have demonstrated for
me that the importance of open standards cannot be underestimated and the best way for
building systems is to adhere to public specifications and promote them where applicable.
The topic of this thesis, the methods of communication (interoperability) of Geographical
Information Systems, is very important in the field of geomatics, and more important, it
affects the life of most practical GIS users. Even more, the enhanced interoperability that
is emerging from the new standards may increase the usage of geospatial data, bringing
new people to the world of GIS by enabling them easily supplement their databases with
locational data. The importance of spatial information and Geographical Information Sys-
tems are constantly increasing, sometimes they are even viewed as the keys to the future of
democracy [49].
1
http://www.r-systems.ee/
CHAPTER 1. INTRODUCTION 2

1.1 Goals
The purpose of this thesis is to give an overview of the field of interoperability of Geographical
Information Systems, especially the development of the latest standards. I would also like
to demonstrate some practical aspects of applying these standards, how they change the life
of both GIS software developers and end users.
The main goals of my work are:

• Analyzing the problems of interoperability in GIS, appropriate standards and practices.


Two methods—files and services are covered.

• Introducing to the wider audience some newly available technologies in the field of
communications of Geographical Information Systems.

• Showing some ways for practically using open standards for increasing the interoper-
ability and availability of GIS services. For this purpose special software products are
created and combined with existing programs. These undertakings are among the first
projects in Estonia that make use of OpenGIS web services.

1.2 Composition
The thesis is composed of seven chapters, here I outline of the main points from each of
them, starting from chapter 2, where a short introduction to GIS world is made with an
emphasis to the problems that are discussed in the following chapters. Additional topics
include a short presentation of metadata and standardization of geodata.
In the 3th chapter two methods of communication and interoperability in the world of
GIS are analyzed. The first part of the chapter covers file exchange, several formats that
are currently used for exchanging geographical data are described and compared. The rest
of chapter 3 concentrates to the web-based communication methods. An overview of such
methods is provided, as well as more detailed description of two OGC specifications, Web
Map Service (WMS) and Web Feature Server (WFS). The chapter ends with conclusions
drawn from the presented material.
The rest of the thesis is devoted to the WMS protocol. Chapter 4 on page 35 presents
several samples of WMS server software. The main focus will be on the architecture of
F-Open server (a product of the R-Süsteemid company, carrying a substantial contributions
by the author of this thesis) but a couple of alternatives are also discussed. The chapter
concludes with comparison of those map servers.
The next chapter (Chapter 5 on page 46) is devoted to WMS client programs. Several
clients for different computing platforms are described and compared.
Chapter 6 looks into some possibilities of using the WMS standard for solving problems
that arose from practical necessities. Three projects are described and the implications of
engaging the WMS standard are highlighted.
Final conclusions are drawn and the key points are reiterated in the 7th chapter. Dir-
ections for theoretical and practical work for the future are provided in that chapter as
well.
The expected audience of chapters 4 and 5 are developers of GIS systems, while the
content of chapters 3 and 6 should be interesting to both developers and end users alike.
Chapter 2

Concepts of GIS

Geographical Information Systems are widely used in every field of our everyday life, often
we use them without even noticing it. But still the concepts of GIS are not very well
understood, even by computer literate people. The purpose of this chapter is to make the
reader, who may possess a lot of knowledge about information systems, acquainted with
spatial data processing and spatial information systems1 .
A Geographical Information System (GIS) is an information system (‘A mechanism used
for acquiring, filing, storing, and retrieving an organized body of knowledge [4].’) which
contains some spatial information, that is data for phenomena on, above or below the Earth’s
surface. Spatial data can be in many different forms: satellite images of the Earth, census
reports in tabular form or digital vector map of city streets [30]. A Geographical Information
System also uses the spatial component of the data, either for visualization or for spatial
analysis. This property differentiates GIS from other systems, e.g. the information system
of a library may also contain maps (in paper and/or digital forms), but as long as it only
stores the data and does not use it, it is not a Geographical Information System.
It has to be noted that nobody has been successful to provide an universal definition of
GIS. The reason is quite simple—similar to other complex phenomena, the GIS has many
variations and is used for many different purposes, giving people disparate impressions about
the essence of a GIS (what parts and operations are important, what components does
it contain, etc.). Some possible definitions together with the discussion on the topic are
provided in the article by D. J. Maguire ([32]).
The outline of a sample GIS is shown in Figure 2.1 on the following page. From this
figure one can see the complexity of a system—operating such systems definitely needs the
knowledge from many domains. Real systems differ from this example, either in hardware,
e.g. they are lacking a digitizer or have an added speech synthesizer, or in software, for
example instead of a Database Management System (DBMS) they use operating system’s
file service. In this paper I will cover only the software side of GIS and will concentrate to
the data storage and interfacing questions as well as to some important properties of the
data (both spatial and attribute data).
1
Some authors prefer to use the term spatial instead of geographical, as the concepts and properties also
apply to other fields where co-ordinates on multiple axes (not necessarily related to the Earth) are used (e.g.
component placing on electrical circuit boards). In this paper I use these words as synonyms because I will
focus on Earth-related spatial data.
CHAPTER 2. CONCEPTS OF GIS 4

Graphics Workstations

Scanner Plotter

Interface
Spatial
Data

DBMS Graphics
System
Digitiser
Attribute
Data Operating System

Computer Communications

Figure 2.1: The physical components of a GIS [2].

2.1 GIS Data Models


The single most important component of every GIS is its data. A GIS usually stores two
different kinds of data: spatial data and attribute data (aspatial data). The difference
between them lies in interpretation, the physical representation can be absolutely the same.
Attribute data is often stored in a DBMS and treated in the same way as in other information
systems (used for making queries and reports) while spatial data is used for visualization or
spatial operations—spatial queries, spatial analysis and other operations. Note that these
two kinds of data are not separated (at least from the viewpoint of a user) but usually
tightly integrated. Typically systems have capabilities for querying attribute data, based on
location (spatial filtering) or vice versa, selecting only the pieces from spatial data that have
some specific attribute values.
Spatial data can be presented in many forms but all digital representations2 can be
divided into two fundamental classes—raster or vector data.
This division originates from our everyday cognition of the space around us (the naı̈ve
geography)—sometimes we look to the objects that fill the space and have spatial attributes,
other times we like to pinpoint a location in the space and enumerate the objects (or more
precisely, the attributes of space) that we found there. These cognitive methods are used
interchangeably, depending on the properties of the objects (discrete, continuous, relative
size, etc.), our experience and other factors. The first approach can be formalized as a vector
model and the second as a raster model [11, 15].
Raster model (raster data) is defined as a tessellation of a plane into regular cells (pixels)3 ,
each one representing a value of some quantity in given location. The whole dataset rep-
resents the field of given quantity varying across the space, spatial position of each cell is
implicit in the ordering of the pixels. The value of the cells of certain areas may be irrel-
evant to our goal but in the raster dataset the pixels are present (and having some values)
2
Here I do not consider narrative forms for storing spatial data as they are not practical in most cases.
3
Raster data is therefore sometimes called grid-cell data.
CHAPTER 2. CONCEPTS OF GIS 5

Property Raster model Vector model


Data entry Faster Slower
Data structures Simple Complex
Positional accuracy Limited by the size of pixels Unlimited
Attributes Good for continuous attribute Good for intrinsic attributes of
fields map features
Data analysis Good for spatial analysis and Good for spatial inquiries
modelling
Storage required Larger Smaller
Transformations of Problematic Easy
co-ordinates

Table 2.1: Comparison of raster and vector models [27].

nevertheless. The cells are usually square-shaped, but in principle they can be any uniform
shape that completely covers the plane without gaps, such as triangles, hexagons or some
fractal shapes [30, 45]. Raster data plays important rôle in remote sensing applications and
other systems where the entities are continuous in space, like a temperature field.
Vector model, on the other hand, contains individual objects (features) which usually
represent real-world phenomena. The spatial components of each object are composed of
points, lines, areas (polygons) and volumes. Only the co-ordinates of special points (end-
points of lines, corners of polygons, etc.) are stored in the dataset, making it more compact
than raster data—no co-ordinate information is kept for a location where no objects are
found. The spatial part of vector data can be treated as a graph with all properties and
operations that are defined for graphs also applying for vector data. Objects have usually
attributes. An attribute is a chunk of character data associated with given object (each
attribute value is similar to the value of a pixel in raster model). An object can possess
many attributes, giving the user the ability to perform various queries on the data [2].
One important property that can be present in a vector dataset is topology. Topo-
logy is generally defined as the spatial relationships between adjacent features, usually in
a two-dimensional space. In GIS, topology is implemented through the structure of data.
Topological data structures are advantageous because they provide an automated way for
handling digitizing and editing errors and artifacts, and reduce the storage that is needed for
polygons because boundaries between neighbouring polygons are stored only once. Spatial
relationships such as adjacency, connectivity and containment, as well as spatial analysis
methods are much easier to implement on topological data structures [54]. Sometimes to-
pology moves from secondary quality of a dataset to the focus, there exist tasks (like the
mapping of principal schema of bus lines of a city) for which the topological relations are
more important than co-ordinates and topological space is used instead of Euclidean space
for plotting such maps (i.e. only topology is shared between the map and the reality, not the
properties of metric spaces (distance)) [18].
Topology has different implications for raster data, as the uniform picture elements of a
raster image cannot support the same set of topological relations that we know for vector
model (or from everyday life). However, a hybrid raster model is proposed, obtained by
completing the raster cell with edges and nodes on its boundaries into a hybrid cell complex.
Such hybrid raster systems support traditional topological relations in Euclidean space and
can be used as a basis for hybrid Geographical Information Systems where (from the user’s
point of view) the data is not divided between vector and raster models [60].
CHAPTER 2. CONCEPTS OF GIS 6

Level Abstraction Description


0 Reality The real world phenomena as they exist.
1 Conceptual model Components and relationships, pertaining to the specific
phenomena thought to be relevant. This data model
is independent of specific systems or data structures.
2 Logical model The logical organization of the data and the manner
in which relationships between components are defined.
3 Physical model The physical layout of the data in computer’s memory
(file structure).

Table 2.2: Levels of data abstraction in GIS [45].

The choice of the data model should be based on the problem to be solved, the intrinsic
properties of the data, the capabilities of the computer soft- and hardware, and the experi-
ence of the personnel. Some comparison points between the vector and raster models, shown
in Table 2.1 on the page before should help to make the right choice. The focus in this thesis
will be on vector data.
The main purpose of every Geographical Information System is to model the reality. The
real world is incredibly complex and is not well suited for data processing. Digital spatial
data always represents a certain model of the real world. The conceptual data model provides
necessary simplifications and regularity for storing and processing the data. The data model
contains nearly always several levels, each level refining and regulating the previous one.
Various abstraction schemas have been proposed for spatial data models, Table 2.2 describes
one such schema.

2.2 Co-ordinates
The location of spatial data is expressed using the co-ordinates. The co-ordinate system
defines the rules for transforming the real-world (Earth) co-ordinates into map co-ordinates
and vice versa. The values of co-ordinates of a point are meaningful only when we know
the datum that was used for determining the co-ordinate values. A geodetic datum defines
the reference system that describes the size and shape of the Earth. There are many co-
ordinate systems in use for different purposes, most are modelling the shape of the Earth
(the geoid ) as an ellipsoid (a sphere is good enough approximation for small scale mapping
and planar model can be used for very large scale surveying). Many ellipsoidal models of
the Earth have been defined during the history, most of which are meant for using only in
particular region. The advent of satellite navigation systems has increased the employment
of geocentric datums (which are usable for all points on the Earth), most notably the World
Geodetic System (WGS) 84 [51].
There are many types of co-ordinates used for mapping purposes. Choosing the right
system may be challenging but it will result in smaller size, better handling, and simpler
processing of the data [34]. The three most commonly used types are shown in Figure 2.2 on
the next page.
The geodetic co-ordinates (geographic co-ordinates) (λ, ϕ) of a point (Figure 2.2, a) are
defined by the ellipsoidal co-ordinates of the projection of the point onto the surface of a
reference ellipsoid along the normal of that ellipsoid. Geodetic latitude ϕ is defined as the
inclination of the normal to the equatorial plane of the ellipsoid. The longitude λ is the
CHAPTER 2. CONCEPTS OF GIS 7

Normal
Zero Zero
meridian meridian PZ

P P
φ Equator PX Equator
λ (0,0,0)

PY

X Y

(a) Geodetic P = (λ, ϕ). (b) Geocentric Cartesian P = (X, Y, Z).

PY P

(0,0) PX X
(c) Cartographic P = (X, Y ).

Figure 2.2: Co-ordinate systems.

angle between the meridional plane (a plane through the normal and the minor axis of the
reference ellipsoid) of the point and the zero meridian [39].
The three-dimensional geocentric Cartesian co-ordinates (Figure 2.2, b) with the origin
at the centre of the reference ellipsoid are sometimes used (A common need for such system
is during the transformation of co-ordinates from one datum to another). The Z-axis is
taken perpendicular to the equator, X-axis is directed along the equatorial plane to the zero
meridian. The Y-axis is perpendicular to the other two. With Earth-centered co-ordinates
we can use the traditional mathematical apparatus for Cartesian co-ordinates and need not
worry about the angular entities of the ellipsoidal co-ordinates. The downside of using this
system is the inconvenience of computing the co-ordinates—changing the position or height
of the observer involves computation of new values for all three co-ordinates (the values of
geodetic co-ordinates do not change when the height is varied). Nevertheless, all satellite-
based navigation systems use the three-dimensional Cartesian co-ordinates internally.
The cartographic co-ordinates (also planar co-ordinates, grid co-ordinates, Figure 2.2, c)
are Cartesian co-ordinates on a plane that are obtained by transforming the co-ordinates from
the three-dimensional space onto the plane using a map projection (analytical) formulae or
numerical methods (polynomial transformations). Two-dimensional co-ordinates are much
CHAPTER 2. CONCEPTS OF GIS 8

better displayed and managed on planar media (such as paper or computer screen). The
big problem with this approach is the impossibility of distortion-less representation of the
Earth’s surface (therefore the distance, measured with ruler on a planar map would be
incorrect in most cases). There exist map projections which preserve some properties of
the ellipsoidal Earth, for example area, local angles, scale, or direction. Comprehensive
explanations of many map projections together with appropriate formulae are provided by
J. P. Snyder ([50]).
The L-EST co-ordinate system which is used by the Estonian Basic Map and most
other spatial databases in Estonia is based on the Lambert Conformal Conic projection with
two standard parallels (58◦ 000 and 59◦ 200 ), European reference frame EUREF-89 and the
GRS-80 ellipsoid [3].

2.3 Metadata
One very important aspect of GIS, sometimes unfortunately underrated in importance, is
metadata. Metadata (metainformation) is generally defined as data about data, although in
some contexts or applications it may have slightly different (usually more specific) meaning.
The supplementary and descriptive data is distinguished from the actual data (the data being
described) only by its content. Metadata provides an architecture, describing a dataset (the
raw data) within a data environment [44].
Spatial datasets are generally large and their structure tends to be complex (especially
when using vector data model) and quite different from one database to another, therefore
the importance of metadata is even greater for geographical data, comparing to completely
aspatial databases.
Metadata can be divided into several categories ([44]):
Access metadata The logical structure of the data and other information needed for quer-
ies and retrieval. Definitions of object classes, attributes, and attribute values. Also
the information about co-ordinate system and relevant parameters.

Semantic metadata The purpose of the data, feature classification guidelines, decision
rules regarding the representation of spatial and aspatial data, spatial integrity con-
straints, etc.

Quality metadata The qualitative parameters that characterize the dataset. For spatial
databases it includes the accuracy (spatial and temporal) of data, time of collection
(or source if compiled from previous work) and other parameters.

Transfer metadata The explanation of the data transfer between applications.

Storage metadata Description of the storage management system (how and where the
data is stored, how often archived, etc.). These parameters are very useful for systems
which contain huge amounts of data on various storage media.
There may be additional categories for providing additional explanations of the data.
Metadata (especially the parts which define the structure of the database, feature classes,
attributes, etc.) is often called data dictionary [30]. Several spatial data exchange stand-
ards (e.g. S-57 and DIGEST) define data dictionaries that represent the conceptual models
(describing the abstractions of the real world) for the respective applications.
The rapidly increasing amount of data that is being produced has led us to the point
where using the metadata is essential for gaining an overview of the available data and for
being able to select interesting pieces from the huge pile of raw data.
CHAPTER 2. CONCEPTS OF GIS 9

Some spatial data storage and interchange formats (see Chapter 3.1) provide means
for storing metainformation together with raw data. Such formats provide usually a pre-
defined set of fields for storing metadata which may be incapable of storing e.g. additional
organization-wide data. Also some datasets may be stored in forms that do not provide such
fields. Therefore an organization-wide metadata database is often needed to keep track of
the various datasets and their qualities. One such system is described in a paper by Michael
J. Meitner et al. ([37]). Such databases should store metainformation in a strict, machine-
processable format (there should not be possible to enter ‘TM’ into the projection field for
one dataset and ‘Transverse Mercator’ for another, provided that the projection is the same
for both). Additional, free-form comments should also be encouraged. When the metainfo
database cannot be applied then the descriptive files (preferably in plain text format) should
accompany each spatial dataset.
The Dublin Core Metadata Initiative 4 is an open forum engaged in development and pro-
motion of interoperable online metadata standards and practices. The Dublin Core Metadata
Element Set (http://dublincore.org/documents/dces/) describes various metadata ele-
ments for common data, such as the title, subject, date, but also type, source and relations,
total of 15 elements. Dublin Core is promoting metadata for all kinds of databases, not only
for spatial data. Dublin Core Metadata Elements contain no special information for spatial
content.
Content Standard for Digital Geospatial Metadata (CSDGM)5 is a specification special-
ized to the digital geospatial metadata, created by the Federal Geographic Data Committee
of the United States of America (USA). The objectives of this recommendation are to provide
a common terminology and definitions for documentation of digital geospatial data. The
standard includes descriptions for various elements in several categories: identification, data
quality, data organization, spatial references, entities (features) and attributes, data dis-
tribution, citations, temporal characteristics, metadata reference and contacts [13]. The
extensive set of metadata elements defined by this standard should provide adequate for
most GIS projects.
The ISO/TC 211 is developing a standard 15046-15 (Geographic information—Metadata)
for specifying a common schema for metainformation for spatial datasets.

2.3.1 Importance of Metadata: An Example


Co-ordinates (either in 2D or 3D space) do not describe spatial data absolutely. Without
knowing the origin, directions, and units of the co-ordinate axes6 the actual values of co-
ordinates are meaningless.
When dealing with geographical data in cartographic co-ordinates we also need to take
into account different projections and datums, as the co-ordinate values for any given feature
depend on the method of mapping the geographical co-ordinates into planar ones.
Figure 2.3 shows two maps of the same area but using different co-ordinate systems,
layered on top of each other. When features have to be copied from one of those maps
(database) to another their locations would be incorrect, unless co-ordinates are transformed
to the datum and projection of the destination map. The appropriate transformations can
only be performed when there is precise knowledge about the source and destination co-
ordinate systems.
4
http://dublincore.org/
5
http://www.fgdc.gov/metadata/contstan.html
6
Assuming Cartesian co-ordinates. Similar requirements can be made for other types of co-ordinate
systems.
CHAPTER 2. CONCEPTS OF GIS 10

Figure 2.3: Two maps of the same area with different co-ordinate systems set on top of each
other. The lower map is in the L-EST co-ordinate system (Lambert Conformal Conic projec-
tion), upper map uses the TM-Baltic co-ordinate system (Transverse Mercator projection).
Scale 1:22000, co-ordinate shift is about 50 m.
Northern coast of Muhu island, Estonian Nautical Chart sheet no. 513.

The importance of metainformation being accompanied with any dataset is clearly per-
ceptible from this example, as we cannot interpret the values of co-ordinates otherwise. The
interoperability of different systems (data sources) is also virtually impossible without valid
formal description (metadata) about the data sources.

2.4 Standardization
The standardization of spatial data and other aspects of GIS has taken off only in recent
years, following the years of de facto standards, set by large corporations. The active
movement for standardization and interoperability (which is virtually impossible without
formal standardization) during the last decade has begun to produce new, high quality
standards that are created by international organizations using the principle of consensus.
Those standards are usually recommendatory but following them is still the only way out of
the quagmire of incompatible systems.
International Organization for Standardization (ISO)7 is the biggest international stand-
ard body whose members are national standard organizations. ISO has approved over
13000 different standards for all domains. International Organization for Standardiza-
7
http://www.iso.ch/
CHAPTER 2. CONCEPTS OF GIS 11

tion has set up a Technical Committee (TC) 211 (Geographic information/Geomatics)8


for standardizing the handling of digital geographical data. This committee has affili-
ations to many other standard-making organizations (International Hydrographic Organ-
ization (IHO), OGC, DGIWG and others) for establishing uniformity and interoperability
of standards.
ISO/TC 211 has a plan to produce over thirty standards, most of which are still un-
released drafts. The topics of those standards range from the spatial model to the mul-
timodal location based services and cover the whole range of problems in the handling and
processing geospatial data. Selected topics include data quality, terminology, spatial and
temporal schemas, different spatial referencing systems, qualification of personnel, and web
map server.
There are also other ISO committees dealing with spatial data, e.g. the Geographic
Data File (GDF) standard, created by the ISO TC/204 (Transport information and control
systems).
The Open GIS Consortium (OGC)9 is an international organization, founded in 1994,
aimed at developing open standards for geoprocessing technologies and growing interoper-
ability in Geographical Information Systems. The members of the OGC are companies,
universities and research organizations around the world that participate in a consensus
process.
OGC provides an abstract specification which contains a framework and reference model
for implementation specifications. It is a high-level guide, defining the theoretical data mod-
els, principles, etc. Implementation specifications are papers that clarify and specify in detail
the interfaces of some part of the abstract specifications. The implementation specifications
are written for software developers. Open GIS Consortium may also provide recommendation
papers which present consortium’s stand on several topics. Recommendation papers are usu-
ally produced as preliminary versions of standards and may develop into an implementation
specification or into a part of the OpenGIS abstract specification.
The implementation specifications include several standards that define interfaces for
providing ubiquitous processing of geographical data in different environments. The ex-
amples of OGC services include map server, feature server, geoparser, gazetteer, and others.
OGC services are examined in greater detail in section 3.2 on page 25. All OGC standards
(both approved and candidate specifications) are publicly available from the consortium’s
web page.
The process of creating and approving an OGC standard is significantly faster process
than approving ISO standards. The fast development cycle, especially at the beginning, and
the open nature of the standardization process help the software vendors to implement the
proposed specifications and providing valuable feedback to the standards committee. The
end users also benefit from the open process, as they can use the specifications and software
implementations of the preliminary versions of the standard to gradually adopt and refine a
particular application of the standard.
ISO and OGC are not the only organizations that have undertaken the efforts of specify-
ing and standardizing some aspects of geospatial data processing. The other players usually
operate on national level (e.g. the Spatial Data Transfer Standard (SDTS), developed by US
Geological Survey (USGS)10 ) or, in case of international organizations, they co-ordinate the
activities in a strictly specialized field (International Hydrographic Organization (IHO)11
8
http://www.isotc211.org/
9
http://www.opengis.org/
10
http://www.usgs.gov/
11
http://www.iho.shom.fr/
CHAPTER 2. CONCEPTS OF GIS 12

the creator of S-57 (IHO Transfer Standard for Digital Hydrographic Data) and many other
standards for hydrography, is a good example of such organizations).
Smaller projects build also on top of the shoulders of ‘giants’ (OGC, ISO, etc.), such
as the GIS Interoperability Project Stimulating the Industry in Europe (GIPSIE) project,
stimulating the interoperability in European GIS industry and seeking answers to some
questions that are not very important to the OpenGIS Consortium (whose members are
mostly from the USA) but are essential in Europe, for example multi-language support and
interoperability in national level [61].
Chapter 3

Interoperability

The previous chapter showed the great complexity and diversity in the field of Geographical
Information Systems. The variety of different systems has risen the question of connectivity
and interoperability between systems. This chapter gives an overview of the needs and
forms of communications in the GIS field. Then a more detailed analysis is done upon some
published standards with the emphasis on file exchange formats and web-based services.
The term interoperability is used to denote co-operation or communication. The standard
ISO 19104 Geographic Information—Terminology defines the interoperability as:

‘The capability to communicate, execute programs, or transfer data among vari-


ous functional units in a manner that requires the user to have little or no know-
ledge of the unique characteristics of those units [19].’

The hardware of modern computers is quite unified and poses few problems—almost
all computers use 8-bit bytes, have the ability of reproduce graphics, etc., the software
layers that reside ‘on top’ of hardware (operating systems, compatibility libraries) provide
additional levelling. Perhaps the most notable problem that arises from the hardware of
computers is the endianness—different ordering of bytes in memory words. The machines
with big-endian processors (e.g. Sun Scalable Processor Architecture (SPARC)) have trouble
reading binary files, written on computers with little-endian processors (like Intel Pentium)1 .
The trouble only affects binary files, text files can be transformed from one to another
without any problems. Of course, the variety of peripheral devices is much greater, some-
times leading to problems when a device cannot be used with some particular computer’s
hardware.
In general, one should not have many interoperability problems with today’s hardware.
This is not exactly the situation in GIS software and from now on the focus in this chapter
will be implicitly on software of Geographical Information Systems.
Although GIS has been along for several tens of years the problems of interoperability in
GIS software had not been regulated by international standards up to now. Data was usually
exchanged in proprietary file formats, some of which gained the status of de facto stand-
ard (for example shapefiles from ESRI or design files (dgn) from Intergraph/Microstation),
regardless on the fact whether the specifications were publicly available or not. In larger
projects seamless work-flow was assured only by careful choosing (or writing new) of soft-
ware components, perhaps from only one vendor. Some projects did so good work that parts
of their systems became de facto standards, e.g. the Topologically Integrated Geographic
1
The problem is easily solvable in software, one needs only know whether the file record is written in big-
endian byte order (memory words are stored most significant byte first) or little-endian byte order (memory
words are stored least significant byte first) and swap bytes when necessary.
CHAPTER 3. INTEROPERABILITY 14

Encoding And Referencing System (TIGER) project of the United States’ Census Bureau
[55].
In this work two aspects of interoperability are covered. First, an overview of existing
GIS file formats is provided, then there are outlined several services providing spatial data
in one form or another.
While the file exchange will undoubtedly remain very important part of data accessibil-
ity processes in Geographical Information Systems, services provide many new possibilities
for data transfer. Services in this context are protocols, which make possible data sharing
between the server and clients. Such services usually build one or more layers upon existing
accessibility protocols2 —Common Object Resource Broker Architecture (CORBA), Com-
ponent Object Model (COM), Hypertext Transfer Protocol (HTTP), etc. Files are usually
distributed as copies of the master database, services, on the other hand, provide the access
directly to the master database. Services are also much more flexible in implementing access
restrictions, have clear advance in cases of frequent updates (Central server has always the
newest data, sending files to all users can create a situation where someone (unintentionally)
uses older data.) and can take GIS into wider audience (interacting with spatial services’
clients in given DCP is usually quite simple). On the downside they usually require on-line
network connections (sometimes considerable amount of network bandwidth is also needed)
and are somewhat harder to implement.
I would like to stress once more the importance of open standards, defined by interna-
tional standard bodies. These standards are based on the consensus of all participants and,
if all parties play fair, all three major groups benefit from such standards: governments
(regulators can back the ‘common goods’ for their communities), companies (providers have
more opportunities for promoting their products and services) and individuals (users are
not locked to systems of only one vendor). Today, every technology provider (company)
can participate in most open standardization processes, either directly in consortia (e.g.
World Wide Web Consortium (W3C), OGC), or indirectly, through appropriate national
standard body, in organizations like ISO. Formally adopted standards also help to prevent
the abuse of the power by the creators of the de facto standards (such as the enforcement
by Unisys of the Lempel Ziv Welch (LZW) data compression patent used in Graphic In-
terchange Format (GIF) images), as participants often have to identify their intellectual
property claims during the standardization process [6, 26].
The need for open standards is perceived not only in GIS community. For example, the
electronics design companies, which, very much like GIS system vendors, rely on intellec-
tual talent and resources, are co-operating in developing new, open standards that benefit
everybody in the industry [16]. They feel that only open standard with wide (unlimited)
range of participants will result in the best, most comprehensive specification. No man is an
island and no company can claim all the knowledge for producing a generic solution. The
situation exaggerates in the field of communication, diverse set of participants result in a
standard that works well with many systems and provides greatly increased interoperability
and efficiency.
Those internationally adopted standards provide a great base for national standards (or
specifications for a specific application field) after necessary customization [21].

3.1 File Exchange


There are lots of different file formats developed for GIS data over the years. Here is
given only a short sampling of some spatial data transfer file formats with the emphasis of
2
Also called Distributed Computing Platform (DCP) [59].
CHAPTER 3. INTEROPERABILITY 15

differences between them. Only vector file formats are under investigation in this section
(although some of the mentioned formats also support raster data model).
Historically, quite a lot of spatial content has been created with Computer Aided Design
(CAD) systems. Although there are similarities between GIS and CAD systems there are
also differences that prevent using CAD software in geomatics (and vice versa). The main
reason is that the GIS data is essentially different from typical CAD drawing. The data used
in Geographical Information Systems is composed of high volume of irregular components of
small number of primitives, often having fractal structure, while CAD drawings contain lot of
regular structures of many (often parametric) types, resulting in a much smaller dataset. The
primary data capture methods are also dramatically different (digitizing versus drafting) [40].
Following sections give an overview of some common exchange formats for spatial data.
Environmental Systems Research Institute (ESRI) shapefile is a minimalist GIS format that
does not support topological relations between objects nor does it store metadata about the
dataset. MapInfo data files (tab and mif/mid files) are also instances of this category.
MicroStation dgn is an example of a CAD file format, Another CAD file format that is
also used for maps and GIS data is Drawing Interchange File (DXF) from AutoCAD.
S-57 and Digital Geographic Information Exchange Standard (DIGEST) represent the
exhaustive spatial data formats that can store a wide variety of geographical data. These
formats support topology, metadata, data dictionaries, and other important elements. Both
formats are created by international standard organizations and are aligned to the same
structures and content when possible. From similar entities SDTS can be mentioned (actu-
ally both S-57 and DIGEST are built using the knowledge from the creation and using of
SDTS format).
Geography Markup Language (GML) is an exemplar of new, Extensible Markup Lan-
guage (XML)-based languages that work well with many XML-aware applications for spatial
and aspatial processing.

3.1.1 ESRI Shapefile


Shapefile is a spatial data format developed in ESRI, originally designed for ArcView 2 in
the early 1990s. The specifications of ESRI shapefiles are freely available from the ESRI
Internet site3 [12, 54].
A dataset in shapefile format consists of at least three files: a main file (data.shp) where
the actual data is stored, an index file (data.shx ) and a database (data.dbf ) for feature at-
tributes. The database is stored as a table in dBASE format. Index file contains offsets of
each feature in main data file, thus allowing random access to data. The dataset may accom-
modate additional files (data.sbn, data.prj and others), which contain additional information
about the data (e.g. the co-ordinate system information is held in data.prj ).
Shapefiles do not contain any hints about the visual representation of data in files.
The main file of ESRI shapefile is composed of a header and variable length data records,
each representing one map feature (shape). The header contains information about the
dataset—shapefile specification version (currently 1000), bounding box, number of records,
and the type of records. One shapefile can currently contain only one type of features.
Specified shape types are [12]:
Null Shape A shape with no geometry. Null shapes can be used in the same file with other
types of shapes.
Point Simple point in two-dimensional plane, represented by its X and Y co-ordinates. All
co-ordinates in shapefiles are double-precision floating point numbers.
3
http://www.esri.com/library/whitepapers/pdfs/shapefile.pdf
CHAPTER 3. INTEROPERABILITY 16

PolyLine A polyline is an ordered set of vertices, representing a polyline (zigzag). Polyline


can have multiple parts (a part is an ordered set of two or more points). Polyline is
located in the XY plane.

Polygon A polygon is a two-dimensional set of one or more rings. A ring is a connected


sequence of four or more points, forming a closed loop (no self-intersections are al-
lowed). A polygon in shapefile may contain multiple outer rings with each one having
possibly ‘islands’, represented by inner rings. The interior of the polygon is defined
by the ordering of points in rings (clockwise for outer and counterclockwise for inner
rings).

MultiPoint A set of points which are not connected. Each point is represented by its X
and Y co-ordinates.

PointM A ‘measured point’—point with X and Y co-ordinates, having also the measure
M.

PolyLineM A polyline with measure attached to each vertex.

PolygonM A polygon with measure attached to each vertex.

MultiPointM An unordered set of measured points.

PointZ A point in three-dimensional space (X, Y, and Z co-ordinates) with the measure M
(total of four parameters).

PolyLineZ A polyline with measures in three-dimensional space.

PolygonZ A polygon with measures in three-dimensional space.

MultiPointZ An unordered set of points in three-dimensional space, each having a measure


M attached.

MultiPatch A Multipatch is a set of patches (surfaces). A patch inside a multipatch can


be either a triangle (several patches form either a strip or fan) or a regular shapefile
polygon. Each vertex has three co-ordinates and a measure M.

It is not possible to store topological information in shapefiles. However, it is possible


to compute the adjacency of shapes and build topology for a dataset if certain prerequisites
are met (e.g. there are no gaps or overlaps between polygons which should be adjacent).
The absence of topological information and relative simplicity, compared to other GIS file
formats (including ESRI’s own coverage format), gives usually faster access times for reading
data from files [54].
Every feature (shape) in an ESRI shapefile can have associated attribute data, which is
stored in the dBASE file (one row for every feature, identified by unique feature identifier).
Attributes can have various data types (strings, integers, floating point numbers, etc.),
according to dBASE file specification [5].
ESRI shapefiles are nowadays read and written by many applications, including all major
GIS programs, and the format has gained popularity, becoming one of the de facto standards
for spatial data interchange. Its simple structure is a bonus in cases when topology is not
needed, but the lack of accompanying metadata would lead to not to recommend its usage
between systems when there is slightest possibility of misunderstanding.
CHAPTER 3. INTEROPERABILITY 17

3.1.2 Intergraph/Microstation Design File


The Intergraph Standard File Format (ISFF) (V7 ) is a CAD file format which is also quite
popular for storing geographical data, especially in the case the data originates from a
CAD-based GIS. The ISFF (also known as the dgn format, as the ISFF filenames have
usually a suffix ‘.dgn’ appended) is the file format common to MicroStation4 and Intergraph’s
Interactive Graphics Design System [38].
The dgn file was designed primarily for storing CAD drawings and its history is clearly
visible from the file structure. ISFF are binary files, either design files (common suffix (file-
name extension) ‘.dgn’) or cell libraries (suffix ‘.cel’). Design files contain both graphical and
non-graphical (invisible) elements. Cell libraries contain a series of cells—complex elements
used for (multiple) placing into design files.
A graphical element is a representation of some constellations (lines, ellipses, etc.) by
Cartesian co-ordinates (co-ordinates are stored as 32-bit integer numbers) and can be a
simple (e.g. line or cone) or complex element (a conglomerate of several simple elements).
Both 2D and 3D elements are allowed. Each element in ISFF is specified to be on one of
the 63 layers. Some of the graphical elements:

Line A line segment, represented by two points.

Line String Three or more connected, ordered points, forming a line string (zigzag).

Point String is a set of ordered points which are not connected (previous points define
orientation at each vertex).

Shape is polygon, formed by a closed line string (the first and last points of the series must
be coincident)

Curve Parametric spline curve, defined by a set of points. The curve passes through all
these points.

B-spline Curves and surfaces defined by B-splines (instead of linear interpolation between
vertices). Design files can contain rational, non-rational, uniform and non-uniform
B-splines.

Text Text with fine-tunable parameters.

Complex Chain is a chain (closed line-string), formed from several simple elements.

Complex Shape is a shape, formed from several simple elements.

Ellipse An ellipse is defined by its major and minor axes, center point co-ordinates and
rotation.

Arc An arc of an ellipse.

3D Surface is a complex 3D element that is projected or rotated from a planar boundary


element (line, line string, curve, arc, or ellipse). Surface is capped at both ends,
enclosing a volume.

3D Solid is similar to surface, but it is not capped, thus enclosing no volume.

Cone A circular cone, truncated cone or cylinder, possibly skewed.


4
MicroStation is a technical CAD system, a product of Bentley Systems, Inc.
CHAPTER 3. INTEROPERABILITY 18

Raster Data This element defines a piece of raster data to be inserted into the design.

The file format from MicroStation V8, released in 2000, is a substantially different file
format, although the files carry the same suffix (.dgn). The new file structure leverages
many problems with the old one, such as the the allowance of infinite number of levels [7].
The specifications of this format are not published.
Every graphical element in ISFF carries information about its presentation. This in-
formation includes the color index (actual colors that correspond to indices (total of 255)
are also defined in the design file), weight (thickness of lines), style (solid, dashed, etc.)
and several different properties. Symbols are usually constructed from the same elements
as other drawings and are stored as cells (complex elements) either in design file or in a
separate cell library [57].
ISFF does not contain, similarly to the shapefiles, any means of storing explicit topolo-
gical relations in the data file. The most one can do with dgn files is to ensure the correctness
of data (that means, for example, that coincident lines are really coincident in all points)
just before the process of data exchange, because spatial datasets are very fragile and editing
is error-prone without the help of topology.
The design file format does not let the user to store metadata in the file, thus severely
crippling data exchange using ISFF (There is a place for storing co-ordinate units and
user can make linkages to outside databases, but using these linkages result in an system-
specific behaviour or total mess in the worst case.). This fact together with the lack of
clean separation between the data itself and its representation leads to the conclusion that
ISFF (dgn) is not appropriate file format for digital spatial data exchange between different
systems.

3.1.3 S-57
The S-57 (IHO Special Publication No. 57) is a standard to be used for exchange of digital
hydrographic data (also known as Electronic Navigational Chart (ENC)) between national
mariners, hydrographic offices, and other users [25].
The full title of the S-57 is IHO Transfer Standard for Digital Hydrographic Data, the
effective version of this specification from November 2000 is Edition 3.1. The text of the
standard is available from IHO for 150 ¤ [24].
S-57, originally developed as a profile of the SDTS, called DX-90, is an official IHO
standard for nautical charts, all hydrographic offices worldwide are required to produce
digital charts in this format.
S-57 is intended purely for data transfer, real applications are expected to convert the
ENC data into vendor-specific (System ENC (SENC)) format that is better suited for dis-
playing and other tasks in GIS (differences between ENC and SENC formats are usually
syntactical, the meaning of data is the same). This is in contrast with previously described
proprietary file formats which were used mostly as a ‘data store’ for certain applications
(The ability of using these files for data exchange was an included benefit that was probably
not a high priority item in the head of the creators of those formats.).
SENC data is most often used in specialized GIS called Electronic Chart Display and
Information System (ECDIS). ECDIS is a complex system, installed on the bridges of ships,
that helps mates in navigating the vessel. The system displays electronic charts with the
data from several sensors (position, speed, surroundings, etc.) and miscellaneous supple-
mentary data (e.g. official notes to mariners and captain’s own notes). S-57 has no concept
of presentation, leaving the visualization of chart content to the applications. In ECDISs
the presentation model defined by S-52—IHO Specification for Chart Content and Display
CHAPTER 3. INTEROPERABILITY 19

of ECDIS is used (other applications can use different presentation models). If the ECDIS
follows the International Maritime Organization (IMO) Performance Standards for ECDIS
then it can be used in place of traditional (paper) charts (The use of S-57 and S-52 standards
is mandatory by the IMO specification) [1, 23, 48].
As a transport form for digital data, the creators S-57 have paid a lot of attention to the
process of updating nautical charts. Updates are given in the form of differences against the
base chart, i.e. there are only inserted, deleted or modified records in the ENC update file.
Those arrangements make easy updating possible even over slow communication channels
(like radio links on board of ships). Checksums are used for all datasets to ensure the
integrity of data during the transfer.
Metadata is quite well handled, each chart has a fixed structure in its heading with fields
for data producer information, parameters of the dataset (accuracy, source, topological level,
and others) and geodetic data (used projection, datum, co-ordinate units, bounding box co-
ordinates, etc.).
The S-57 data model is composed of feature objects and spatial objects. Feature object
(feature) is abstraction of a real-world phenomenon, it contains descriptive attributes but
has no geometry. Feature objects are located by spatial objects. Each feature object has an
associated geometric primitive, indicating whether the object’s type is point, line, or area
(objects which do not reference any geometry are also possible). Areas may have insets
(‘islands’). Similar feature objects belong to the same class (e.g. roads or wrecks). Classes
are defined in the data dictionary.
Spatial object contain the geometry, although it may also contain descriptive attributes.
Vector geometry is currently the only available form of spatial information that can be coded
in S-57.
All co-ordinates are given in two- or three-dimensional space. Three-dimensional co-
ordinates are only used for sounding points (feature class SOUNDG)5 , for all other objects the
third dimension is expressed, when necessary, as an attribute of an object (e.g. the height
of a lighthouse). Locations of points can be expressed in planar or in geodetic co-ordinates.
Vertices are connected either by straight lines, elliptical arcs, Bezier, or B-splines.
The S-57 format can store topological information and therefore it is often not possible
to describe the geometry as one spatial object but the geometry of one feature object have
to be broken up to ‘bits and pieces’ with additional relations between them.
A spatial object is either a isolated node or connected node (representing a point), an
edge (representing a line), or face (representing an area). There are four topological models
defined in the standard which use different relations on different types of spatial objects for
encoding the geometry:

Spaghetti is a data structure where all lines and points are unrelated to each other (no
topological relations are stored). Edges and isolated nodes are used, areas are repres-
ented by edges that track the perimeter of the polygon.

Chain-node is a geometrical data structure, composed of edges, isolated nodes and connec-
ted nodes. Edges and connected nodes are topologically linked. Lines are composed
of edges which reference a connected node at each end, areas are represented by edges
that track the perimeter of the polygon.

Planar graph is a two-dimensional data structure in which the geometry is described in


terms of nodes and edges which are topologically linked. It is a special case of a chain-
5
Soundings have also exceptional feature object type—they may have an arbitrary number of unrelated
co-ordinate points (multipoint geometric primitive), while formally being point objects.
CHAPTER 3. INTEROPERABILITY 20

node structure where edges are not allowed to cross (a connected node is inserted to
all junctions).
Full topology is a two-dimensional data structure in which the geometry is described in
terms of nodes, edges and faces which are all topologically linked. A planar graph with
faces.
The connection between feature and spatial objects depends on the topological model
used: for spaghetti model there is one-to-many relationship, any objects with coincident geo-
metry (like land area and coastline) results in duplication of spatial data. Other topologies
do not allow the duplication of co-ordinates and many-to-many relations must be used. It
is also possible to define feature collections (relationships between features).
One important part of the S-57 standard is the data dictionary. Data dictionary is part of
the metadata for ENC, it describes feature classes, attributes of features and spatial objects
and relations between objects and attributes (which attributes are allowed for each object
class). A class has a name, type (geographic, cartographic, meta, etc.), numerical code,
short and long descriptions. Attributes have also name, code, descriptions, but also type
(integral, real, string, enumerated) and domain (a set of permitted values, specified either
by enumeration or by boundaries). Data dictionary is transferred from data providers to
users like other datasets (either in full length or as updates), and is then installed into user’s
GIS (ECDIS or other). Individual ENC datasets do not include the data dictionary.
S-57 uses the ISO/IEC 8211 standard (Specification for a data descriptive file for inform-
ation interchange) for encapsulating digital hydrographic data. That standard provides a
mechanism for transferring digital data (in the form of files) from one computer system to
another, independent the make and characteristics of such systems, or the medium used for
transmission.
ISO/IEC 8211 uses so called data descriptive records for describing the logical structure
of the actual data contained in the file. With such schema it is possible to decode the file
with no a priori knowledge about the structure of the data file (field names, lengths and
data types).
S-57 has an appendix called ENC product specification which refines and clarifies the
main part of the standard. This appendix also imposes major restrictions to all conforming
S-57 datasets: all charts must use chain-node topology with linear interpolation between the
points, all points must be in geodetic co-ordinates (using the WGS 84 datum), textual data
has to be in English, and so on [25].

3.1.4 DIGEST
The Digital Geographic Information Exchange Standard (DIGEST) is developed by DGIWG.
The most current version of this standard is edition 2.1 from September 2000. DIGEST is
an open standard, its text is freely available to everybody at the DIGEST Internet site6 .
The purpose of this standard is enabling the seamless transfer of digital geographic inform-
ation between Geographical Information Systems. DIGEST supports the exchange of both
raster and vector data. Standard specifies data structures, means of data encapsulation and
the data dictionary (Feature Attribute Coding Catalogue (FACC)) used for describing the
dataset [9].
The DIGEST format resembles, at the conceptual level, the United States’ Spatial Data
Transfer Standard (SDTS). It is built in close co-operation and alignment with other inter-
national standards, such as IHO S-57, a suite of geomatics standards developed by ISO/TC
211, etc.
6
http://www.digest.org/
CHAPTER 3. INTEROPERABILITY 21

The dataset can be encapsulated in a number of ways: ISO 8211, generally used for
bulk transfer and for archival purposes, ISO 8824 (Abstract Syntax Notation number One
(ASN.1)) used in telecommunication, Vector Relational Format (VRF) for database user
applications, and Image Interchange File (IIF) for image and raster data. DIGEST defines
also media and labeling standards for magnetic tape, optical discs, and other exchange
media. In addition, file naming conventions and character representation are also specified.
Datasets that are formed in accordance with DIGEST guidelines can contain topological
information. DIGEST supports vector data with four levels of topology and also raster data.
The topological model and levels are identical to those used in S-57 (Minimum Redundant
Topology). Vector features are classified as point, line, area, text, or complex (containing
other features) features. Spatial extent of features is expressed in node, edge, and face ele-
ments (these elements can carry positional attributes (e.g. about the measurement accuracy
of co-ordinates of a given point)). Points are specified in three-dimensional space. Raster
data files contain pixels whose value (colour) is defined as a red-green-blue triplet or by using
look-up tables for getting the correct colour value. Co-ordinates for registering the raster
image are also stored with the image.
Information about attributes and feature classes is kept in a data dictionary (Feature
Attribute Coding Catalogue (FACC)). Each feature object is identified by a code which also
identifies its category (culture, vegetation, etc.).
Extensive metadata is included with vector and raster datasets, containing data source
information, projection and reference points for scanned raster maps, various data quality
indicators, such as completeness and consistency levels, accuracy, and much more.
The DIGEST specification has been customized to form a national standard for cadastral
data [21].

3.1.5 GML
Geography Markup Language (GML) is an XML encoding for the transport and storage
of geographic information, including both the spatial and aspatial properties of geographic
features. The GML specification (current version is 2.1.1 from 25 April 2002) is prepared by
the Open GIS Consortium and can be found at the OGC site7 . Current status of the GML
standard is an OGC Implementation Specification.
The purpose of the GML standard is to provide an open, vendor-neutral framework for
the definition, storage and transport of spatial application schemas and objects for specialized
domains, to encourage the sharing of application schemas and the information they describe.
Geography Markup Language is designed to separate the content cleanly from its present-
ation and the standard only specifies the encoding of spatial data without any notices to the
visual representation. The wide range of XML tools can be used to transform the data in
GML to any visual form (raster or vector graphics, voice, etc.) [20].
GML has great impacts on the systems that deliver spatial data over the World Wide
Web (WWW). Typically, maps in the WWW environment used to be synonyms to raster
images (GIF, Portable Network Graphics (PNG), etc.). GML brings a new dimension into
the spatial data exchange via Internet with the separation of the content of the map cleanly
from its presentation. The visual rendering is done on user’s side, with user-specified tools
and parameters. Also, maps, encoded in GML are ‘true’ (vector) maps, supporting queries
and editing, enlarging the possible audience. There are also an increasing number of tools
that can be used for processing GML data as the XML technologies make steady progress
in the WWW environment [17].
7
http://www.opengis.net/gml/02-009/GML2-11.html
CHAPTER 3. INTEROPERABILITY 22

A GML document is described in a metalanguage called XML Schema [62]. The current
GML standard specifies three base schemas: Geometry schema for encoding geometry, Fea-
ture schema for describing basic properties of features, and XLink schema for making links
between objects, these three are normally referenced in user-defined schemas. Users of GML
are expected to create new application schemas for their application domains, based on the
three schemas provided by OGC.
The standard allows (and encourages) the creation of profiles—subsets of the GML stand-
ard that fit some specific application domain. An application schema is then created from
the basis of a given profile.
The data model of GML is based on the OGC Abstract Specification 8 [52] which provides
the reference model for Open GIS Consortium Implementation Specifications. A dataset
which represents the real world is defined by a set of features, each having properties (at-
tributes). Some properties of geographic features describe its geometry. Features can be
aggregated into feature collections.
GML operates with with the simplification of the OGC Abstract Specification called
simple features where the geometric properties of the features are restricted to ‘simple’
geometries for which co-ordinates are defined in two dimensions9 and the delineation of a
curve is subject to linear interpolation.
The Feature schema defines a small set of geometric properties for association of geomet-
ries (defined in the Geometry schema) with features. Examples of these properties include
location, centerLineOf, and others.
The Geometry schema describes the primitive geometrical constructs that form the basis
for spatial databases. The current version supports only vector data (this restriction results
from the simple features specification). All co-ordinates can be specified in two- or three-
dimensional space. Each element may have an optional pointer to a co-ordinate reference
system. All elements of geometry collections (MultiPoint, MultiLineString, MultiPolygon,
Multigeometry) must share the same spatial reference system, which is included only for the
collection element. The Geometry schema defines following geometrical elements ([20]):

Point element represents a single co-ordinate pair (triple).

Box is a rectangle, represented by two co-ordinate tuples (the minimum and maximum
values along co-ordinate axes). Box elements are used for denoting extents.

LineString is a piece-wise linear path defined by a list of at least two co-ordinates that are
assumed to be connected by straight lines.

LinearRing is a closed LineString where the last co-ordinate tuple must be coincident with
the first one. LinearRings are used for building Polygons.

Polygon is an element representing a connected surface, made up from LinearRings. A


Polygon can contain ‘islands’ (holes) inside the main area.

MultiPoint is a collection of Points.

MultiLineString is a set of LineStrings.

MultiPolygon is a collection of Polygons.


8
http://www.opengis.org/techno/abstract.htm
9
GML 2.1 includes some rudimentary support for three dimensions (co-ordinates of a point can be specified
in 3D space) but no geometrical constructs are allowed to use the third co-ordinate.
CHAPTER 3. INTEROPERABILITY 23

Property SHP ISFF DIGEST S-57 GML


Metadata no no yes yes yes
Co-ord. system yes* no yes yes yes
File type binary binary binary binary text
No. of files many varies one one varies
Topology no no yes yes no
Presentation no yes no no no
Company
yes yes no no no
controlled
Specifications free rev. eng free non-free free

For explanations see the text of section 3.1.6.


Table 3.1: Comparison of selected GIS file formats.

MultiGeometry is a container that can hold arbitrary geometries (primitive and/or col-
lection elements).

Version 2.0 of GML does not support topology, although future versions are expected
to have this capability. The base schemas provide very few hooks for attaching metadata
to the spatial dataset (e.g. names of the features, spatial reference systems for geometries).
At first glance it may seem not appropriate for today’s systems where metadata is given
an ever increasing rôle. When one dives deeper into the GML standard it becomes clear
that by the creation of application schemas everyone can adopt any particular dataset to
have exactly the amount of additional (meta) information one believes sufficient. GML with
accompanying schema definitions is perhaps the most flexible format available today for
encoding spatial data.

3.1.6 Comparison of File Formats


Different spatial data exchange formats possess quite diverse characteristics as they are
created with different tasks in mind. If one is about to choose the file exchange format
for spatial data he may look at some crucial properties of file formats that were covered in
previous sections, as summarized in Table 3.1.
The rest of this section contains explanations to the table.

Metadata Whether the format supports metadata (remarks about the main data) in the
dataset.
ESRI shapefiles and ISFF files do not provide any means for storing metadata in the
data files, users must put the accompanying metadata (e.g. data source and accuracy)
to free-format files and take care of distributing them together with data files.
DIGEST and S-57 provide predefined fields for storing metadata. This indeed encour-
ages the using of descriptive (meta-) data but unfortunately also restricts the user to
predefined set of values. However, in most cases the standardized form for metadata
is sufficient.
The most flexible format is GML which puts the full weight to the shoulders of im-
plementors of a profile (sub-format). The GML specification describes only the co-
ordinate system metadata (see next item), the profiles are free to define and use
whatever descriptive data is needed for given applications.
CHAPTER 3. INTEROPERABILITY 24

Co-ord. system The possibility to attach co-ordinate system info (positional metadata)
to actual files. This kind of metadata is essential for interpreting the co-ordinates
correctly. Unfortunately the DGN files do not contain the place for such information.
ESRI shapefiles have a place (prj file) for storing co-ordinate system data but a large
number of datasets do not include this information (this is a later extension to the file
format specification).
File type The type of files, either binary or text. GML, like all XML formats, is a text-
based file format while the other formats covered here apply binary encoding on the
files. Text files have their advantages and disadvantages, the main benefit being the
readability by humans as well as by computers, providing the ability to check or modify
the file ‘on-the-fly’ using only a text editor.
On the other hand, they take a lot more space on storage media and require more
processor power for parsing than their binary counterparts. For example, a simple
GML file, containing only one polygon feature with 177 vertices on the outer ring and
7 on the inner ring, and having 13 attributes takes as much as 20 KB of storage. The
binary representation of the same data would need a fraction of this space.
No. of files Number of files in typical dataset (here we do not count in the data dictionaries,
presentation libraries, style-sheets and other auxiliary files). S-57 and DIGEST provide
the whole dataset in one file while the shapefiles require the division into multiple files
(The restriction of only one type of shapes in one file, one may also want to separate
different layers (with different attributes) into separate files.).
The DGN files support all graphical elements in one file so the dataset can be stored
in one file. Sometimes the dataset is broken into number of files according the layers
of the drawing or element types.
The GML does not make any assumptions about the number of files and leaves the
problem to the users. While it is possible (and often most practical) to put all data
into one file it is also legal to break the dataset into smaller pieces (GML files tend to
grow quickly in size). Referencing between those pieces is supported.
Topology indicates the the ability of storing topological relationships between features in
given format. S-57 and DIGEST are the only files from the that support topology.
Storing the geometry in topology-aware data structures should be essential for all non-
trivial spatial databases and the support of topology should be considered mandatory
for formats, used for data exchange for ‘solid’ and ‘serious’ tasks.
It is important to notice the different topological models that can be used in these
files. Other data exchange formats not covered here are known to use yet different
models which may complicate the data reuse (conversion between topological models
is needed).
Presentation info Whether the information about the visual appearance of data (line
width, symbol shapes, etc.) is described in the same file. Only the ISFF stores
presentation integrated into the data file (like other CAD systems do). Such tight
coupling makes the change of presentation very painful.
The opposite practice of making separate files for instructions that translate spatial
data into meaningful visual appearance is preferred in most situations, as such present-
ation file can be used for many datasets and they can be changed quickly in case
different symbolization is needed.
CHAPTER 3. INTEROPERABILITY 25

Company controlled indicates whether the file format is controlled by one corporation
or not. Clearly the data exchange formats that are not driven by one company but
an international standard body or consortium are preferred, as the latter include the
wisdom from larger number of people and are generally designed to interoperate better
with different software products.
The individual corporations (however big they are!) tend to be more volatile than
consortiums and either stop supporting the products that handle the specific data
formats or even cease to exist, leaving its customers dead in the water.
Specifications Whether the specifications of given file format are available gratis (free)
or for a nominal fee (non-free). The specifications of the proprietary file formats are
often not available (to wider audience) at all. This used to be the case with DGN
format but its specification is disclosed lately ([38]). However, this specification does
not provide full description of all the fields in the file, the meaning of which is being
reverse engineered by volunteers [56].

3.2 Services
The ISO standard on the terminology of geomatics gives the following definition for the term
service:

‘A capability which a service provider entity makes available to service user entity
at the interface between those entities [19].’

It can be seen from this description that the service provider and the service user work
together for the process of data exchange and users get on-line access to the data. This
is in contrast to file exchange where the data source (provider), more often than not, is
independent from the data consumers and work is done off-line.
Services are implemented on top of communication protocols (frameworks) which imple-
ment the low-level primitives for connecting two applications. In distributed environments
the communicating processes can reside anywhere in the network. Popular platforms used
nowadays include HTTP (the foundation behind the now-ubiquitous World Wide Web),
CORBA (a broker-based distributed environment), .NET (a new generation XML-based
DCP) and others.
Some frameworks (e.g. CORBA) provide broker services for locating service providers or
consumers, while others (e.g. HTTP) do not and the task of locating clients, servers or peers
in the network is left completely to users of the protocol.
A client can connect to many services at the same time using different communication
frameworks, as shown in Figure 3.1 (technically, the client application on the figure has to
implement two separate client interfaces for the two DCPs).
The purpose of this section is to give an overview of some services that provide spatial
context to their users. Due to the great variance in available services and the limited space
only web-based services (using the HTTP for data transfer) are discussed here.
Most attention is given to the standards developed by Open GIS Consortium. These
specifications represent the first attempt to standardize the field of spatial web services. As
the OpenGIS standards are open and well-thought, they will surely get wide adoption in the
nearer future. When the OpenGIS infrastructure is universally accepted it becomes easy to
implement various types of services providing geospatial content that plug seamlessly into
the network of other similar services and become instantly accessible to the users.
CHAPTER 3. INTEROPERABILITY 26

service service service


provider provider provider
1 2 3

DCP 2
DCP 1

client

Figure 3.1: The overall concept of services.

Standardized web services will lead us to unification, increased interoperability, and


wider utilization of spatial information and services. Using web services for accessing geo-
spatial data can, in longer run, provide a whole environment for miscellaneous web-based
Geographical Information Systems—WebGIS. A WebGIS is a complex system with access
to the Internet for capturing, storing, integrating, manipulating, analyzing and displaying
data related to locations without the need of having any proprietary GIS software [43].

3.2.1 Proprietary Map Servers


The first web site that included geographical maps was established by Xerox Palo Alto
Research Center in 1993 [46]. Today most big GIS software vendors have developed some
forms of web services (usually in a form of a map server which creates raster images based
on the data layers and location, specified by the client.). There are wide variety of such
servers (see, e.g. an article by F. Limp ([31]) for comparison of various map servers) that
differ in functionality, methods of user interaction (client interface) and price. Unfortunately
there is no interaction between those web services, a user cannot combine the results (e.g.
raster images of maps) of different services.
An example of a proprietary map server is Maptech MapServer (Not to be confused
with the MapServer from the University of Minnesota, covered in section 4.1.), which can be
found at http://mapserver.maptech.com/homepage/index.cfm. The server provides map
images for locations that can be specified by co-ordinates or, using the geocoding capabilities
of the server, by postal codes or city names. The geocoder works only for the addresses inside
the USA, co-ordinates can be specified for all other places as well. Four kinds of maps are
provided: topographical, nautical, aeronautical and aerophotos (not all of these are available
for all regions). Maps are rendered as Joint Photographic Expert Group (JPEG) images.
The client side is a simple application, composed of Javascript and Hypertext Markup
Language (HTML) that runs in a web browser. Zooming (a few predefined values) and
panning are supported, geographical co-ordinates are provided for map center. An interesting
CHAPTER 3. INTEROPERABILITY 27

feature of Maptech MapServer is the ability to insert and label private points of interest on
the map. The label is composed of a text (possibly containing a hyperlink) and a symbol
from a large array of predefined symbols.
Standards are emerging to remedy the previous ‘disastrous’ state of the World Wide
Spatial Web [22]. Existing services often try to give as many attraction as possible, stuffing
many functions into one interface (Maptech MapServer) without clean separation between
them (the output of a geocode is displayed as a map, without any possibility to get explicit
co-ordinate information, for example). Formal standards, developed by OGC (covered in
the rest of this chapter), specify clean interfaces for every functional unit (e.g. geocoder,
co-ordinate transformation server, map server, etc.) that are able to work together but are
perfectly usable individually.

3.2.2 OpenGIS Web Services


The OpenGIS Consortium, being an international industry-wide organization, has created
a set of high quality specifications for different types of spatial services. This section gives
an overview of the OpenGIS Web Services and outlines the common components of these
services.
The OGC Web Services are divided into two groups: OpenGIS Web Mapping Services
and OpenGIS Geospatial Fusion Services. OpenGIS Web Mapping Services include stand-
ards that make possible the dynamical querying, accessing and processing of spatial inform-
ation in the WWW environment. The parts of the Web Mapping Services family include
WMS, WFS, Web Registry Service (WRS), and Web Coverage Server (WCS).
OpenGIS Geospatial Fusion Services deal with the non-map information that refer to
places10 . Such information can be in various formats: text, video, photographs, and so
on. These services support different activities, such as searching, discovery and sharing of
spatial information, standards of this group handle different aspects of spatial information
depending on the content and purpose of the data (addresses, place descriptions, etc.).
The Geospatial Fusion Services include Gazetteer Service, Geocoder Service and Location
Organizer Folder and others [41].
This work concentrates to the first group of standards (Web Mapping Services), a longer
overview is of these and other OpenGIS standards is included in the paper by Kuus, Krusberg
and Rebane ([29]). The next sections describe some selected standards from the OGC Web
Mapping Services.

Common Elements in OpenGIS Web Services


This division discusses some bits and pieces that are common to all web services as specified
by Open GIS Consortium.
All OGC web services use the HTTP as an underlying transport protocol, there are
also expansion hooks for adding support for other DCPs in the future. The communication
between the server and client is carried out by requests, which are sent by the client to
the server using platform-specific techniques (e.g. HTTP GET method). The server then
returns the appropriate answer (map image, grid coverage, error message, etc.) after it has
finished processing the request. Each request contains a number of parameters whose names
and values are defined in appropriate standards. Software vendors may include additional
vendor-specific parameters which introduce additional possibilities to the systems that are
aware of them (usually the systems of the same software vendor).
10
I.e. the spatial information that is not expressed explicitly in co-ordinates but in some other form instead,
like a postal address.
CHAPTER 3. INTEROPERABILITY 28

Every compliant web service has to support the GetCapabilities request which returns
description about the service, its information content, a list of supported protocols, requests
and parameters. This information is provided to the clients of the service in XML form,
providing easy parsing by computers as well as by humans. One physical server may be
able to simultaneously act in place of several logical services (e.g. WMS and WFS), in this
case the server capabilities document will contain information about all these services. The
GetCapabilities request is supposed to be the first contact between a server and a client in
the session. During this request the version negotiation is carried out when necessary.
All OGC web services support step-by step negotiation of version of a service protocol
that will be used between the service provider and user. The process of negotiation is
carried out as a part of a GetCapabilities request: if the version in the client’s request is not
supported by the server it responds with another version number which is either accepted
or rejected by the client. In the iterative process of negotiation the server and the client will
come to an agreement over the exact version of the protocol that will be used for further
communication (if they cannot find a common subset of the protocol the following actions
would be meaningless).
Whenever a server encounters a situation where it cannot recover from (i.e. invalid para-
meter value supplied by the user), it throws an exception back to the client where the request
came for. Common XML-based exception formats are defined for all OGC services, there
may be other formats available for reporting errors (e.g. WMS standard defines exceptions
in the form of raster images). Client can specify a preferred exception format in its requests.

3.2.3 OpenGIS Web Map Service


The Web Map Service (WMS) is an Open GIS Consortium standard for a service that
produces georeferenced maps in the WWW environment. The full title of this normative
document is Web Map Service Implementation Specification and it is a OpenGIS Implement-
ation Specification, currently at version 1.1.1 [59]. The specifications are available for free
download on the OGC web site11 .
The WMS standard is the most mature of all OGC web services, it also formed the basis
for the ISO standard ISO/WD 19128 Geographic information—Web Map Server interface
(currently in the status of a working draft), being developed by the ISO/TC 211 (Geographic
information/Geomatics). The ISO standard is meant to be compatible with the OpenGIS
Consortium specification.
WMS builds on top of the HTTP protocol and uses a HTTP server as a mediator between
WMS server and its clients (Figure 3.2 on the following page).
The WMS protocol specifies three requests: GetCapabilities, GetMap, and GetFeature-
Info. An example of a session between the server and a client is shown in Figure 3.3 on the
next page (real clients would probably make several GetMap and GetFeatureInfo requests
during a session).
An example of the WMS GetMap request for drawing three layers: http://195.50.203.
246/F-Open-server/server.php?VERSION=1.1.0&REQUEST=GetMap&LAYERS=Base-NC610,
Navigation-NC610,Lights-NC610&STYLES=&SRS=EPSG:3300&BBOX=551122.3,6592254.9,
556126.4,6599329.6&WIDTH=406&HEIGHT=574&FORMAT=image/png.
In the context of WMS the map is defined as a visual representation of geodata, a map
is not the data itself. This specification defines three WMS operations (requests): GetCap-
abilities returns service-level metadata, which is a description of the service’s information
content and acceptable request parameters, GetMap returns a map image whose geospa-
11
http://www.opengis.org/techno/specs/01-068r3.pdf
CHAPTER 3. INTEROPERABILITY 29

HTTP INTERNET
server

WMS
WMS client
server
Figure 3.2: Data flow in WMS.

GetCapabilities request

WMS_Capabilities document

GetMap request

map image

GetFeatureInfo request

GML document

client server
Figure 3.3: A sample session between a WMS server and a client.
CHAPTER 3. INTEROPERABILITY 30

tial and dimensional parameters are well-defined, GetFeatureInfo returns information about
particular features shown on a map. The implementation of the GetFeatureInfo request is
optional to WMS software and service providers [59].
The data in WMS servers is divided into layers. A layer is an atomic entity for the user
request (GetMap). Some layers may be marked as queryable, in this case the clients are
entitled to make GetFeatureInfo request for these layers. For each layer there are defined
the name of the layer (which is used in client’s requests), spatial extent (bounding box), a
list of Spatial Reference Systems12 , and other parameters, such as additional dimensions,
units, pointers to additional information about the layer’s data, etc.
For each layer there have to be one or more styles that define the presentation of layer’s
content. Normally the WMS servers use static styles that can be referenced by their re-
spective names. It is also possible to use WMS servers together with the Styled Layer
Description (SLD) specification that provide the user-defined symbolization for map layers.
With SLD the client gets a map image that is visualized using the user-supplied data, result-
ing in the desired visual appearance of the map [10]. The SLD WMS adds some additional
requests to the protocol that are not covered here.
The map image produced by a WMS server is created with well-defined geospatial and
dimensional parameters and can be used wherever there is a need for spatial data. Two or
more maps can be blended into a composite image when they have the same bounding box,
Spatial Reference System (SRS), and dimensions.
The most common form for the map image, generated by the WMS server, is some
ubiquitous raster image format—Tag Image File Format (TIFF), PNG, JPEG, etc., but the
image can be also in vector graphics format (e.g. Scalable Vector Graphics (SVG)) or in
non-graphical descriptive form (GML), although the latter is being largely superseded by
the Web Feature Servers. Other data that is occasionally sent from server to its clients is in
XML form, making the automatic parsing and understanding it easy for the client programs.
A WMS server can act as a cascading server, that is act as an WMS client to other servers
while still performing as an WMS server for its clients. Cascading server has to redirect some
(or all) of the requests to other servers and send the output returned by those servers back
to its client. Such intermediating servers can also perform various ‘enhancements’ on the
data, such as image format or co-ordinate system conversions, or aggregation of data from
several other servers.
The WMS server throws exceptions, as all OpenGIS web services, when it cannot further
proceed the current query. In addition to the standard OGC XML-based exception format
WMS also defines exception indication messages in the form of raster images. Either a blank
image or an image with the text describing the situation is sent instead of map image in
case of a raised exception.
The requests defined in the WMS specifications [59]:
GetCapabilities The GetCapabilities request gives service metadata about the informa-
tional content (technical contacts, access restrictions) and acceptable request para-
meters of the particular server (Uniform Resource Locator (URL)s and file formats).
The information about the data (keywords, spatial and temporal extents, SRS, data
provider contact, etc.) and its presentation (styles) is provided for each layer. The
description is provided in a machine- and human-readable XML form.
GetMap The GetMap operation is designed to produce a map, either as a pictorial image
or as a set of graphical elements. The resulting map is produced from the user-specified
12
The WMS specification uses the co-ordinate systems from the European Petroleum Survey Group (EPSG)
database of geodesy parameters (http://www.epsg.org/) or alternatively by specifying the parameters of a
projection (only a few co-ordiante systems are supported using the latter method).
CHAPTER 3. INTEROPERABILITY 31

list of layers and styles, using the bounding box, SRS, width, and height supplied in
the request parameters.

GetFeatureInfo The GetFeatureInfo request is designed to provide clients of a WMS with


more information about the features in the maps that were returned by previous Get-
Map requests.
The parameters of this request include the parameters of a GetMap request (from this
information the WMS server determines the location and scale of the area of interest).
Additional parameters provide the layer names to be queried, pixel co-ordinates of the
point that caught user’s interest and the desired format of returned information.
The server returns the map features that are found in the specified point. A server can
report map features in several formats, as described in its response to the GetCapabil-
ities request. The GML is often used for this purpose for its fitness for the task (easy
processing, tailorable for different application domains, relatively small data sets). The
information returned by the GetFeatureInfo request resembles that of a WFS (see the
next section).
The GetFeatureInfo request is acceptable only for layers that are marked queryable in
the service metadata document.

Some implementations of WMS servers and clients are covered in Chapters 4 and 5.
Chapter 6 on page 56 gives some examples of utilizing the WMS protocol for solving real-
world problems.

3.2.4 OpenGIS Web Feature Server


While the GetFeatureInfo request (indeed all other requests of the WMS specification too) of
WMS only supports one-way data exchange between server and client—map features are sent
from server to clients for inspection (and possibly for saving to client-side storage) but there
are no methods for the client to alter the data in server’s database. The WFS specification
contains different requests for clients for retrieving the feature data from the server and also
modifying the data in server’s datastore. WFS is meant to supplement WMS, the two can
be coupled very tightly.
The Web Feature Server (WFS) is an OGC standard (Web Feature Server Implementa-
tion Specification ([58])) specifying the handling (querying, updating, insertion, deletion) of
map features in spatial databases using the HTTP protocol.
A WFS request contains a description of query or data transformation operations that
are to be applied to a web-accessible datastore. The WFS specification is independent of
the type of database (Relational Database Management System (RDBMS), XML-files, etc.).
All requests (and server responses) are encapsulated into the HTTP protocol, using either
GET or POST methods. The standard form of data exchanged between the WFS server
and client is XML (GML for spatial data). Exceptions are raised when the server cannot
parse user’s input or other abnormal events occur.
The standard defines two classes of Web Feature Server servers: the Basic WFS, which
would provide read-only access to the datastore, and the Transaction WFS which provides
also full support for transaction-based modifications in the database.
The Basic WFS must support GetCapabilities, DescribeFeatureType and GetFeature in-
terfaces, Transaction WFS must support these, plus additionally LockFeature and Transac-
tion requests. A sample session between a server and a client is sketched in Figure 3.4.
CHAPTER 3. INTEROPERABILITY 32

GetCapabilities request

WFS_Capabilities document

DescribeFeatureType req.

FeatureTypeDescription doc.

GetFeatureWithLock req.

GML document

Transaction request

TransactionResult document

client server
Figure 3.4: A sample session between a WFS server and a client.

Features can be queried either one-by-one or using a filtering system defined in Filter
Encoding Implementation Specification. The OpenGIS standard for filter encoding provides
a flexible system for specifying spatial and aspatial constraints on a set of data [14].
Features in the database are differentiated by a unique identifier. The structure of the
database is addressed by the feature schema (a data dictionary, describing the properties of
a feature type, their data types, constraints on the data types, etc.). The feature schema is
provided by the server using the DescribeFeatureType requestand it must be consistent with
the OGC feature mode, defined in the OGC Abstract Specification ([53]). The knowledge of
the feature schema, used by the server, lets the clients construct features (e.g. for inclusion
into the database) that do not raise structural conflicts and integrate seamlessly to the
datastore. The standard provides means for specifying specific vendor specific extensions
(such as the use of specific Structured Query Language (SQL) constructs on RDBMSs).
The WFS protocol supports the cascading of servers, providing opaque access to sev-
eral datastores for the clients. Cascading servers are capable of performing homogeneous
requests on other servers (all objects of one query are from one datastore). A Web Feature
Server providing distributed service also supports heterogeneous requests, where the objects
of one query may be physically located in different databases (servers). Such services require
advanced locking techniques (two-phase commit) from the servers.
The Web Feature Server standard defines the following interfaces (requests):

GetCapabilities This request describes the capabilities of the Web Feature Server in a spe-
cific XML document that is modelled closely after the WMS capabilities specification.
The capabilities document contain sections about the service metadata—service pro-
CHAPTER 3. INTEROPERABILITY 33

vider contact information, access URLs for requests, relevant keywords, vendor-specific
extensions, feature types with operations defined on them, and other information.

DescribeFeatureType The purpose of this interface is to provide the means for requesting
a definition of any feature type that a particular WFS can serve. The feature schema
(data dictionary) is expressed in the XML Schema ([62]) language13 . The feature
schema should adequately describe all objects that are accepted by that server.

GetFeature The GetFeature interface allows clients to retrieve features from a Simple
Feature datastore. One request can contain multiple queries, selecting features by
their properties (attributes). Client can also instruct the server to lock the features
that are returned as a result of these queries.
The server responds to this request by sending out data in XML format.

LockFeature The LockFeature request is used by Transaction WFS for locking specific
objects in the datastore. The need for explicit locking results from the inherent state-
lessness of web connections which would result in unpleasant side-effects when sev-
eral clients are using the same object for write access (update, delete) simultaneously.
Without locking the results of some of these operations would be lost and the database
could be corrupted. Using the LockFeature request grants an exclusive lock (‘owner-
ship’) for given objects to the client performing this operation. Other transactions
trying to modify the locked object would fail during the lifespan of the lock. Clients
can specify the time-out before the server expires the lock automatically (to avoid
deadlocks), the completion of the Transaction operation on the locked object by the
owner of the lock will also release the lock.
The LockFeature request can operate on multiple objects, using the filtering sys-
tem [14]. The implementation of the locking operation is opaque to the client.

Transaction The Transaction interface is used to describe data transformation operations


that are to be applied to a web accessible datastore. The WFS server translates the
client request into the language of the target datastore and executes the transaction. A
notification is sent to the WFS client about the result of the operation. The identifiers
of the newly created objects are also returned in case of insertion.
The Transaction request can perform Insert, Update, and Delete operations which are
performed on the objects previously locked by the LockFeature interface. Transaction
can operate on a subset of locked features, with the possibility of leaving others in
the locked status. The GML data describing the objects to be modified should be in
accordance to the feature schema.

3.3 Conclusion
Two methods for exchanging data were discussed in the current chapter—files and services.
The specifications of several instances of both methods were covered.
A question rises now, whether to use files or set up a server for exchanging data for a
particular project? The choice should be made individually for each case as it depends on
many factors. Some points to consider are shown in Table 3.2. Although files are traditionally
used for importing and exporting data from applications, services are gaining importance
13
WFS vendors can also support other schema description languages, such as Document Type Definition
(DTD).
CHAPTER 3. INTEROPERABILITY 34

Quality Files Services


Network connection not required mandatory
Latest data probably always
Data discovery complicated easy
Integration with other (non-GIS) systems harder easier
Implementation of access restrictions harder easier
Complexity of implementation lower higher
Usability in legacy systems better worse
Data provider and consumer both active no yes

Table 3.2: Comparison of file and service based interoperability methods.

as data providers. Nowadays the geospatial services should be considered seriously for new
systems, especially for applications that operate in a Distributed Computing Platform (e.g.
the Internet) where there exist plenty of such servers.
If the file based data access method was chosen then the question of file formats needs
to be solved. Table 3.1 on page 23 summarizes the properties the file formats that were
covered earlier in this chapter. Pros and cons may be different for various situations, but
attention should be paid to metadata: skipping it is easy if it is not needed but a dataset is
often unusable due to the lack of metadata. Therefore the appropriate support for metadata
should have high priority whenever a file format is selected. Topology should also considered
essential for large and complicated datasets. From the covered file formats DIGEST is
probably most appropriate for complex data while GML may be preferable for data with
simpler structure and lower volume. GML should also considered for cases when XML
based formats and technologies are used for other (aspatial) data, as the unified processing
of spatial and aspatial data can provide substantial simplification for the whole application
or technological chain.
When services are found superior to file exchange then the question about selecting
an appropriate standard rises. In current situation I would suggest to use relevant OGC
specifications, as they provide the best solutions, especially for web services.
Although some proprietary service protocols may appear more powerful and elegant they
are often inappropriate for actual use, either for lack of accompanying metadata, insufficient
abstraction, or other reasons. Currently available geospatial services are mostly map servers
that provide raster images, generated from spatial databases. OGC has specification for
similar functionality (WMS) but also for many other services.
The model of services as proposed by the Open GIS Consortium is qualified for a large
range of problems, from the very simple ones (e.g. elementary public map servers) to highly
complicated, multi-tier data stores that can accommodate several types of data and provide
appropriate interfaces for accessing the database (e.g. WFS for vector features, Web Coverage
Server for satellite imagery, and so on). The strength of the OGC specifications lies in the
modular structure—all services are perfectly usable separately, yet the true elegance of the
standards is revealed when two or more services are used to complement each other.
The technical excellence (extensive use of XML technology, good support for metadata,
etc.) of the OGC specifications should provide enough motivation for using them instead of
proprietary protocols in most (if not all) cases14 . Open protocols should be supported as
only formal standards allow for the most people to get the most benefit.
14
For products which ‘talk’ multiple protocols, proprietary and open, like ESRI ArcIMS, the open interfaces
should be definitely enabled.
Chapter 4

WMS Servers

An overview of some tools for providing interoperability in the GIS world was given in the
previous chapter. The topics of the next three chapters concentrate to the implementations
of OpenGIS Web Map Service (WMS) specification, providing information about the servers,
clients, and the overall process of solving the real word problems using the WMS.
This chapter analyzes three distinct WMS server implementations. Two map servers,
CubeSERV and MapServer, are discussed briefly in appropriate sections. Most attention
is given to the F-Open server, a WMS server implementation developed in Estonia. The
author of this thesis has contributed substantial parts to the design and implementation
of the F-Open server, which let me provide here a deeper insight into the key points of the
design of this application. The chapter ends with the comparison of the three servers.

4.1 MapServer
MapServer 1 is an on open source development environment for building spatially enabled
Internet applications. The version of the program is currently at 3.5 [35].
The MapServer was originally developed by the University of Minnesota (USA) For-
Net project in co-operation with National Aeronautics and Space Administration (NASA)
and the Minnesota Department of Natural Resources. Additional enhancements have been
funded by several organizations and private companies.
It is a Common Gateway Interface (CGI) application that can be compiled to run on
Linux, Unix and Microsoft Windows systems. The output of the program (a web page)
is generated from template, providing easily configurable look. The outputted page can
contain the generated map image, controls for zooming, panning and querying map features,
and other elements, provided by the owner of the server.
MapServer uses several other open source or free software packages for accomplishing the
task of creating spatial applications: Proj.42 cartographic projections library, GD library for
dynamic creation of images, and many others.
MapServer can load and process spatial data in lots of formats, both vector and raster:
ESRI shapefiles and ESRI ArcSDE vector formats, TIFF/GeoTIFF, GIF, PNG, JPEG, and
other raster formats. Support for these data formats is in the core of MapServer, many
more kinds of data can be utilized by configuring the MapServer to use the Geospatial
1
http://mapserver.gis.umn.edu/
2
http://www.remotesensing.org/proj/
CHAPTER 4. WMS SERVERS 36

Data Abstraction Library (GDAL)3 for raster and OGR Simple Features Library4 for vector
formats (the development of these two libraries is headed by Frank Warmerdam).
The data may be given in any of the vast number of map projections provided by the
Proj.4 library. MapServer can be set up to handle on-the-fly re-projection of the data by
the user request.
The output map image is generated by the GD library, the format of the map raster
image is any of the image types supported by the library.
The visualization of spatial data is carried out by an user-defined symbology. Map
features can be presented by markers (point symbols), lines, or shadesets (area styles).
A particular symbolization is bound to an object by a set of rules, described by regular
expressions that are combined with simple logic operators (conjunction, disjunction and
comparison), on feature’s attributes. The rules can also specify the minimum and maximum
scales where the map object is visible. MapServer can automatically create appropriate
legend for the visualized data, showing the meaning of different graphical elements.
The MapServer system supports MapScript which allows popular scripting languages
such as Python, PHP Hypertext Preprocessor (PHP) and others to access the MapServer
Application Programming Interface. Using MapScript it is easy to create web pages with
maps in everybody’s preferred language. MapScript gives the users full control over the
MapServer program—graphical rendering of spatial data from multiple data sources, map
queries, automated generation of the legend and scalebar, and so on.
MapServer understands OGC WMS versions 1.0.0, 1.0.7, and 1.1.0. It can also handle
the cascading of other WMS servers. In fact, there is no WMS client in the MapServer
distribution, users who want to access WMS datasets are supposed to set up their own
MapServer installation that represent WMS queries in the MapServer CGI interface. In this
case MapServer is acting as a cascading WMS server whose last leg (closest to the user)
speaks its own protocol, not WMS.
Most of the characteristics of MapServer, described earlier, are also available via the
WMS interface, the only differences are in the request parameters and server responses (A
WMS server sends an map image in response to the GetMap request, while the MapServer’s
native behaviour is to output the whole HTML page.).
MapServer understands the GetCapabilities, GetMap, and GetFeatureInfo requests. The
XML response for the GetCapabilities request is generated automatically from the MapServer
configuration file.
GetMap can return maps in several raster formats, provided by the GD library (PNG,
JPEG, or Wireless Bitmap (WBMP)). The result of the GetFeatureInfo query can be re-
quested by the client as plain text, HTML formatted by a template, or as GML.

4.2 CubeSERV
CubeSERV is a WMS server that is developed by CubeWerx, Inc.5 , located in Quebec,
Canada.
CubeSERV is compliant to the latest WMS standard (version 1.1.1 at the moment of this
writing). It supports cascading the services of other WMS servers and all three request types
that are specified in the WMS standard. The server supports data in multiple co-ordinate
systems and several raster image formats (GIF, PNG, and JPEG) for the GetMap request.
3
http://www.remotesensing.org/gdal/index.html
4
http://gdal.velocet.ca/projects/opengis/
5
http://www.cubewerx.com/
CHAPTER 4. WMS SERVERS 37

It can serve dynamic (live) maps from large (and possibly distributed) databases, being is
powered by the CubeSTOR spatial warehouse system.
CubeSTOR is a geospatial warehouse system that provides the storage, management, and
access to very large volumes of spatial data. Another purpose of the geospatial warehouse
is to ensure the uniform data model across the system and keep the appropriate metadata,
catalogues, etc. up to date.
CubeSTOR is based on the Oracle RDBMS system. The system includes tools for creat-
ing and maintaining geospatial warehouse, middle-tier application server, and the presenta-
tion interface (the CubeVIEW application). CubeSTOR is supported on multiple computing
platforms: Windows NT, Linux and several incarnations of Unix [8].
The CubeSERV WMS server license costs ¦8000, additional payments are necessary for
support and maintenance contracts. The cost of the license of the CubeSTOR Geospatial
Warehouse is ¦27000 (this includes the license of CubeSERV).
The publicly accessible demonstration of CubeSERV (which is a great source of in-
teresting map data due to its large number of map layers and cascaded servers) is loc-
ated at http://www.cubewerx.com/demo/cubeserv/cubeserv.cgi. The CubeVIEW pro-
gram, available at http://www.cubewerx.com/wmt/cubeview/cubeview.cgi can be used
as a graphical user interface for browsing the data.

4.3 F-Open Server


F-Open server is a WMS server produced in the R-Süsteemid company, capable of servicing
the WMS GetMap request (it supports several raster image formats for generated maps)
and also able to form GML output for the GetFeatureInfo request.
F-Open server is a commercial grade WMS server that is known to work flawlessly with
wide range of WMS client programs. It accepts requests in WMS protocol versions 1.0.0 and
1.1.0 (the latter is preferred, ‘native’ version), supporting all three request types that are
specified in the Web Map Service standard. A descriptive error message is sent to the client
every time the server is not capable of fulfilling a particular request. Client can specify
the preferred format of error messages (Three formats are available: an XML document,
conforming to the standard exception DTD from OGC, a blank image, or an image with the
textual description of the problem.) and image formats in its requests to the server.
The PNG and JPEG raster formats are currently supported for image responses (gen-
erated maps and exception indicators) while XML is used for textual data returned from
the WMS server to the user. F-Open server can output several types of XML documents—
WMS capability definitions, GML data in response to the GetFeatureInfo request, and error
messages.
The WMS standard specifies the division of data, provided by a map server, into layers.
In case of F-Open server a layer contains the data one or more feature classes from a Frege
map file. It is not possible to use the data from multiple files on one layer, although several
layers can display data from one file.
F-Open server does not limit the number nor size of data sources (map files) for a server
instance, it is limited only by hardware (mostly by memory capacity). Each layer may
have an arbitrary number of associated styles that provide different visual looks (graphical
languages) of the map.
F-Open server is commercial software that is licensed for a fee. The price is negotiated
separately for each customer, depending on the desired functionality (and possible modifica-
tions), the amount of data being served, and the number of concurrent users. F-Open server
CHAPTER 4. WMS SERVERS 38

is also offered by R-Süsteemid as a part of larger packages, which include the WMS server
and customized client software, data file converters, etc.
A working demonstration of F-Open server is located at http://www.r-systems.ee/
F-Open-server/server.php.

4.3.1 History
The history of the project goes back to the beginning of the year 1999 when the R-Süsteemid
company decided to implement an Internet map server based on the Frege map engine. The
first prototype was implemented as a CGI program written in the C programming language.
The program started, loaded a map from hard disk, drew the map requested image and
exited. This was clearly inefficient as it caused many unwanted activities, such as the
initialization of map engine or loading the maps from files to memory. In addition the map
visualizing was performed using the drawing engine of X Window System, Version 11 (X11),
then converted to X Pixmap (XPM) file which was displayed in user’s browser, as the Frege
library could do only X11 drawing those days.
Later that year a client-server model was implemented, consisting of a server (mapserver)
which handled the map loading and drawing parts, and a client, which was a CGI program
brokering user requests to server and presenting a nicely formatted HTML page with the
returned map image to the user. Both programs were written in the C language, they
communicated using a network socket. The server was implemented as a Unix daemon
process which loaded the maps during its initialization phase and fulfilled subsequent requests
by rasterizing the vector maps into images.
The CGI part parsed a template HTML file and when it encountered an item from a set of
proprietary macros inserted into that file it expanded the macro into the appropriate control
(communicating with the server when necessary), resulting in a standard HTML page which
was delivered to client browser. These macros included a set of selectable buttons for map
zooming, a map picture itself, map panning controls and more. This way the appearance of
the map pages could be easily re-designed (by a graphical designer) without the need of any
re-programming.
Around the same time support for several other output formats were added to Frege
using the free GD library6 , developed by Thomas Boutell. The GD library (and therefore
also Frege) supports PNG, JPEG and WBMP images.
Later, multi-threading support was added to mapserver and it became tmapserver. Multi-
threading made the server usable in real-world situations where several requests could arrive
at the same time, that is another request might arrive at the time server is still processing
the first one. Previous versions produced correct output only when the request were sent
sequentially (requests could not be overlapping in time). tmapserver was capable of servicing
ten simultaneous connections.
Other improvements included the rewriting of the client module in the PHP programming
language, partly as an exercise to learn new technologies and partly for making the process
of changing the mapserver front-end easier. The overall architecture (client-parsed macros,
etc.) remained the same.
In the spring of 2001 tmapserver was redesigned to meet the requirements of the OpenGIS
standards, the resulting product was renamed to F-Open server.
6
http://www.boutell.com/gd/
CHAPTER 4. WMS SERVERS 39

MAP

DATA PRESENTATION
DICTIONARY LIBRARY

Frege

DISPLAY USER INTERACTION

Figure 4.1: Handling maps with Frege.

4.3.2 Frege
Frege7 is a library of functions for handling digital vector maps. Frege is a direct descendant
of the Mapper library, described in my diploma project [28]. The author of this work is
one of the principal developers of the map handling library. The Frege library is grown
out from the program code that displayed nautical charts, now it is completely reworked
and its internal structure can accommodate the majority of different spatial data formats
that are in use nowadays. The other strong point besides universality that characterizes
our map engine is performance—the first application that used Frege was a hydrographic
data acquisition software on board of a survey ship where fast map drawing routines were
necessary, as the computers would not have spare processor cycles for collecting data from
several sensors [47].
Frege includes subroutines for loading and saving maps, displaying the maps with arbit-
rary visible area and scale, changing the map content (editing of map objects or metadata),
and making spatial queries on maps. The principal schema of operation is shown in Fig-
ure 4.1.
The inner data model for Frege is derived from the S-57 standard. In addition the
ability to use different kinds of maps (based on different data dictionaries) with different
presentation libraries is added by R-Süsteemid. Maps can use any of the four topology
models, defined in S-57 (spaghetti, chain-node, planar graph, or full topology), all map
editing functions are aware of topology. A rich set of meta-information is also attached to
each dataset, describing the co-ordinate system, dataset parameters (e.g. creation time and
creator), number of objects, etc.
7
The name ‘Frege’ was chosen after the logician G. Frege, to stress the importance of semantics in com-
munication (The Frege’s triangle links together sense, expression and referent.).
CHAPTER 4. WMS SERVERS 40

Maps are stored in a flat files for efficient loading times. All file operations in Frege are
done with these files, data can be imported to Frege format via converters for S-57, dgn,
shapefiles, and MapInfo tab and mif/mid formats. Export of modified map data is possible
to S-57 files.
Frege contains no hard-coded information about the map semantics, it is stored in data
dictionaries instead. Each map is associated with a data dictionary that provides the defin-
itions for all classes of feature objects and a possible set of attributes for each class. With
different data dictionaries Frege is applicable in very different application domains.
Neither data files nor data dictionaries contain no information about the visual style of
map elements. Separate presentation libraries are used for visualizing map content. There
may be more than one presentation library suitable for one data dictionary (a type of maps),
making the alternate visual representations of data very easy, one just need to draw a map
with another presentation library.
Presentation library is a powerful concept for visualizing spatial data. It is a graphical
language that provides look-up tables for matching each map object (feature) to a set of
graphical primitives, based on object’s class, geometric primitive and attribute values during
the map drawing process (see Figure 4.2). A presentation library can possibly contain several
object look-up tables, color tables, and other parameters, which can be selected individually.
These parameters let the end user (e.g. WMS clients) select among the alternative visual
looks for the same set of data, choosing the most characteristic and expressive form, based
on user’s needs.
There are several types of graphical primitives:

• line drawing styles (simple ones (LS ) for variable width coloured lines and complex
line-styles (LC ), where line is visualized by a repetitive pattern).

• area fill instructions, either with one color (AC, four levels of transparency) or with
repetition of pattern symbols (AP ).

• symbols for point locations—SY.

• text drawing instructions with selectable font family, size, and justification (TX, TE ).

• complex symbolization primitives (CS ) provide hooks for external functions (from a
dynamically loadable library) that can provide more complicated symbolizations than
the look-up tables of the presentation library. Complex symbolization procedures can
also take advantage on the user-changeable presentation parameters for providing more
dynamic visualization.

All symbol and pattern definitions are given in a special vector drawing language which is
independent of the final output system.
Two target systems are available for map drawing—X11 (network transparent windowing
system, used mainly on Unix workstations) and the GD library, the latter providing support
for drawing into various raster image file formats. Adding new code for other target systems
is straightforward due to modular design of the output subsystem of Frege.
A work is currently underway for interfacing Frege with a DBMS. PostGIS8 will be used
for a spatial database, feature attribute data would be located in any Open Database Con-
nectivity (ODBC) compatible DBMS. These arrangements take advantage of the powerful
engines of today’s database systems for using them for attribute (aspatial) queries using the
SQL. The Frege-database link is used for the WMS server of the registry of assets of RMK.
8
PostGIS (http://postgis.refractions.net/) is a set of facilities built on top of a PostgreSQL relational
database to achieve the conformance with the OGC Simple Features for SQL specification.
CHAPTER 4. WMS SERVERS 41

MAP OBJECT
class
geometric primitive
attributes

LOOK−UP TABLE

SYMBOLIZATION INSTRUCTION

SY, LS, LC, AC, AP, TX, TE, CS

MAP
DISPLAY

Figure 4.2: The work-flow in the Frege presentation library.

Depending on the results of the current project a continuation project may be following
for allowing also spatial queries to be carried out by the database engine on behalf of Frege.
Using the PostGIS database for spatial data storage also makes it easier to co-operate with
other systems that use the Simple Features for SQL ([42]) standard for accessing geodata.

4.3.3 System Architecture


F-Open server is composed of two parts, front-end system and server proper, as shown in Fig-
ure 4.3 on the next page. These two parts are separate programs that communicate over the
network. Such design gives great flexibility in the configuration, as the server proper and
front-end may run on separate computers. Such separation have positive effects on load bal-
ancing and the overall security of the system. Indeed, only the front-end (with its relatively
small code base) is required to be on the ‘outer’ side of the firewall, the server proper may be
located behind the firewall. All network ports other than the port used for communication
with front-end, may be left closed on the computer that hosts the server proper.
One important aspect of the F-Open server is the image cache. It is a structure that stores
all generated map images together with the parameters (image size, location, map layers,
etc.) of the request that are needed for image generation. A quick look-up is performed
from the cache for every GetMap request that arrives, and if the parameters of the current
request match a cache entry the time-consuming rasterizing phase is bypassed and the map
image from the cache is returned to the client. The image cache has a finite size, if the cache
is full and a new request arrives, the least recently used entry is purged from the cache and
CHAPTER 4. WMS SERVERS 42

MAP MAP
front−
end 1  

client 1

main thread front−


thread 1 end 2
thread 2
thread 3 client 2

front−
end 3

client 3
F−Open server
Figure 4.3: The schematic diagram of the F-Open server.

the result of the new request is stored in its place. The size of the cache is adjustable from
the configuration file.
F-Open server may be instructed to answer to the GetFeatureInfo requests for certain
WMS layers. When the server gets such query it searches through the provided layers and
returns the information about all objects that contain the point of interest, as specified by
the client. The whole set of spatial (co-ordinates) and aspatial (attributes) information that
is known to the server about these objects is returned. The information is supplemented
with the descriptions from the data dictionary (feature class name, attribute names and
values, etc.). The feature data is returned in the GML format, the owners of the server can
specify proper XML schemas and namespaces for each queryable layer.
According to the OpenGIS WMS standard an exception is raised and an error message
is sent back to the client every time something unusual happens. A client can request
exceptions to be reported as a standard XML notation document, as a blank image, or as
an image with the textual description of the message.
CHAPTER 4. WMS SERVERS 43

Front-end
The task of the F-Open server front-end system is to do the initial processing of the para-
meters of the request and forward them to the server proper and to send the results(images
or XML documents) back to the user after the server has completed the processing of the
user request.
The front-end is written in the PHP scripting language and is executed by the Apache9
web server (the interpreter for the PHP language is compiled into the web server). The
web server’s job is to listen to the HTTP requests and start the F-Open server front-end
script when the HTTP request comes to the address of F-Open server (see also Figure 3.2 on
page 29).
The F-Open server front-end then starts to process the WMS request by parsing the
parameters of user inquiry. Besides the formal syntax checking it also does away with
the syntactic differences between versions of the WMS protocol when necessary. For ex-
ample, WMS version 1.0.0 defines the accepted values for the REQUEST parameter as map,
capabilities, and feature info, while the WMS version 1.1.0 specifies the values GetMap,
GetCapabilities, and GetFeatureInfo. This step is needed to keep the communication
protocol between front-end and server proper the same, regardless of the WMS protocol
version specified in the client’s request.
Front-end is also responsible for version negotiation with the client, as specified in the
WMS standard [59]. F-Open server is capable of servicing requests for WMS versions 1.0.0
and 1.1.0.

Server Proper
The server (server proper ) is the part of F-Open server that is responsible for fulfilling the
requests of WMS clients. It interacts only with the F-Open server front-end, not with the
WMS clients.
The server proper is a multi-threaded program (daemon) that waits for requests from
the users (via the front-end system) and spawns a new processing thread for each user
connection (Figure 4.3). All these program threads run in parallel and access the same data
store. No signaling between threads is necessary as the WMS server does not modify the
data (some data structures, such as the image cache, are explicitly locked when write access
is mandated).
The F-Open server proper is implemented in the C programming language, it uses the
libxml2 10 library (created by Daniel Veillard) for parsing and validating XML documents.
F-Open server has a configuration file where the server administrator can specify the ex-
act behaviour of the server instance. The file is an XML document that is basically identical
to the WMS capabilities XML. Some additional elements are introduced for F-Open server-
specific components, such as the names of log file, map files, presentation libraries and para-
meters, and so on. The configuration file is loaded at the start and validated for correctness
against the DTD. This step eliminates the need to check the presence of all mandatory para-
meters and the overall well-formedness of the configuration in server code, removing much
hassle. One can also force the server to reload its configuration by sending the HUP signal to
F-Open server process (killall -HUP fopenserver).
The F-Open server writes traces of all events to the log file. Actions that trigger logging
include normal user requests (client host name, request parameters, etc.), error conditions
and other relevant information. Each record in log file is written together with a time-stamp
9
http://www.apache.org/
10
http://www.xmlsoft.org/
CHAPTER 4. WMS SERVERS 44

and thread identifier, providing additional information for later analysis on server’s usage
patterns and/or tracing the misbehaving of the program.

4.3.4 Future Plans


Plans for the future enhancements for F-Open server include the implementation of WFS
functionality, support for cascading servers, and the ability of dynamic (on the fly) re-
projection of maps. High priority is given for developing more data import converters for
the Frege library.
An important addition for the map server would be the ability to ‘speak’ in different
languages. This improvement would create the possibility for the user of the map server to
choose the language (via client applications) for the dialog with F-Open server, allowing the
internationalization of WMS systems—e.g. the public server of Estonian Maritime Adminis-
tration should be able to communicate in Estonian, English, Finnish, Russian, and Swedish
languages (Most people interested in Estonian nautical charts are using these languages.).
The texts that server sends to a client, meant for human users (layer names, descriptions,
error messages, etc.) will be in the language that the user selected in his request. Available
languages are defined in an extended capabilities XML document.
This idea will be realized as an extension to the WMS protocol, as OGC has not yet
created a specification for multi-language support in its services. The language selecting
option would thus not be available in all WMS client programs.

4.4 Conclusion
This chapter gave an overview of three distinct WMS servers: MapServer, CubeSERV and
F-Open server . A more detailed description of the design and implementation F-Open server
provided some ideas about the techniques used in implementing the OGC WMS specification.
Following list shows some points of comparison between the three WMS servers in-
troduced in this chapter. These arguments are applied here to compare F-Open server,
MapServer, and CubeSERV, but they can be useful for evaluating other map servers as well.

Functionality The most important criterion for anybody selecting a map server would
be the functionality. All three applications implement the OGC specifications and
although each has its small quirks they can all interoperate quite flawlessly. They are
all on par in regard of this property.

Supported spatial data formats Section 3.1 demonstrated the great variance in the spa-
tial data formats. The primary component of every GIS system is the data and there-
fore everyone needs to choose the right tools that can handle the formats of the data
that are needed for a particular case.
MapServer can handle both vector and raster data of different kinds, when MapServer
is built with the OGR and GDAL libraries it can utilize most widely used data formats.
F-Open server does not support many different data formats and cannot handle raster
data at all, but its strength is in the handling of IHO S-57 format that is very important
to all mariners (MapServer can also load S-57 datasets but it does not have proper
support for correct representation of the data, according to the S-52).
There are few facts about the CubeSERV but I suppose it can also load geospatial
data in most common formats.
CHAPTER 4. WMS SERVERS 45

Output formats Another aspect of comparison is the type of output. It is important for
cascading WMS servers and also for end users who may want to save the map images
and use them in other programs.
While the format of the GetCapabilities response is fixed by the OGC the formats of
the responses to the other two requests are not.
F-Open server, MapServer, and CubeSERV all support PNG and JPEG raster image
formats the GetMap request (CubeSERV can additionally produce GIF and MapServer
WBMP images). These image types are the most common in Internet and should pose
no problems.
For the GetFeatureInfo all three support the GML format, MapServer and CubeSERV
can also output features in plain text or HTML format. Again, the GML is the
preferred format by the OGC and should become the standard for spatial data exchange
in Internet.

Cost The total amount of resources (i.e. time, money, knowledge) are nearly always limited
and should be used up most efficiently.
The price of the WMS server sums up from the cost of the WMS server software, other
software that is needed (operating system, database systems, additional libraries, etc.)
and hardware. These sums are one-time investments (until the upgrade of software
or hardware) but there are also running costs associated with map servers. The ex-
penses for actual spatial data should taken into account as well as the cost of network
bandwidth, which may be the largest by far of all expenses for a WMS server.
MapServer is a clear winner in this section as it has no associated costs for software.

Performance The overall performance of a map server depends on many factors starting
from the hardware, the amount and organization of the data, network bandwidth, and
ending with the quality of implementation of the WMS server.
The importance of the performance varies with the purpose of the server, greater
performance is never bad but some critical systems will need good performance at any
cost.
It is nearly impossible to compare the performance of three different systems that
can use various spatial data formats. However, these are some ‘rules of thumb’ for
evaluating the performance of map servers:

• The execution of CGI programs (MapServer) is inherently slower than a separate


daemon (F-Open server) due to the additional cost of allocating resources for the
new process.
• Most GIS software has a ‘native’ file format, using data in this format would
yield the best performance. E.g. MapServer should give the best performance
with shapefiles and F-Open server with S-57 data.
• Generating the map image is the most common operation of the WMS server,
it is often also the most time consuming operation. F-Open server uses its image
cache to leverage this problem—the ready-made image is taken from the cache if
there has been a request with exactly the same set of parameters lately.
Chapter 5

Web Map Service Client Programs

This chapter is dedicated to the client side of the Web Map Service specification. Several
client applications are described in this chapter.
The complexity of the client applications ranges from nil (the person inserts the URLs
manually into the standard web browser) to great (the WMS client is part of a much more
complex application). Also, the applications that provide access to WMS servers often do
not implement the whole specification. Some of them do not support the GetFeatureInfo
interface, some can handle spatial data only in certain co-ordinate systems, some have im-
proper user interface (user cannot choose styles for layers or cannot save the map image, for
example), etc.
There are an increasing number of WMS client applications available, anybody who needs
to get data from a WMS server should choose an appropriate one for his needs. This chapter
shows different ways of implementing client-side applications for the OGC WMS protocol,
programs are chosen to reflect the diversity of methods in various implementations of WMS
clients.

5.1 CubeVIEW
CubeVIEW (Figure 5.1 on the next page) is a data visualization application for the Cube-
STOR data warehouse system from the CubeWerx, Inc. It includes a web-based geodata
navigator that can act as a WMS client program (Described as a template application for
on-the-fly generation of HTML and JavaScript code for various spatial data visualization
frontends). CubeVIEW is a CGI program that runs in the service provider’s server, user
interaction with the program is done using the elements of HTML forms (text fields, buttons,
etc.). CubeVIEW is provided free of charge to the purchasers of CubeSTOR or CubeSERV
systems [8].
CubeWerx has created a web demo of the CubeSERV/CubeVIEW system (works with
all modern browsers), available from http://www.cubewerx.com/wmt/cubeview/cubeview.
cgi. The demo site defaults to CubeWerx server but it is possible to specify other WMS
servers as well.
CubeVIEW provides easy navigation (centering, zooming and panning), manual selection
of scale and center point, image type and size and also other parameters. As it is built
without using the Java language nor the capabilities of the most recent web browsers, it
does not provide the ‘dragging zoombox’ functionality. This is nevertheless hardly noticeable
when actually using the program, as the navigational part is very well thought-out and
intuitive.
CHAPTER 5. WEB MAP SERVICE CLIENT PROGRAMS 47

Figure 5.1: CubeVIEW, a WMS client interface from the CubeWerx, Inc.
CHAPTER 5. WEB MAP SERVICE CLIENT PROGRAMS 48

User can easily add or remove layers from the list of visible maps, using hierarchical
selections. Presentation styles can be selected separately for all layers.
The user interface contains many more elements in addition to the WMS map image
and navigation controls. The most visible is the world map with a rectangle (or point,
depending on scale), showing the area of the main map. There is also legend of visible layers
(if supported by server(s)), descriptions of the server and activated layers are available behind
a hyperlink.
CubeVIEW supports large number of co-ordinate systems but unfortunately not the L-
EST (EPSG:3300) system, widely used in Estonia. Therefore it is not possible to use the
CubeVIEW interface for accessing the majority of Estonian geospatial databases.

5.2 WMSClient
WMSClient (Figure 5.2 on the following page), a flexible WMS client application, is a
product of the R-Süsteemid company. It is implemented in the Java programming language
(needs Java version 1.3.1 or higher), which gives platform-independence for practical pur-
poses. Indeed, one can supposedly execute this program on any computing platform where
there is an implementation of the Java Virtual Machine (JVM) (there are JVMs for all major
operating systems nowadays). WMSClient has been verified to work without modifications
on Microsoft Windows 98 and Linux operating systems.
WMSClient can run as a stand-alone program or as an applet inside the web browser. In
either case it runs in the local machine of the user (as opposed to the CGI approach) which
improves the power and quality of available tools and widgets (especially for user interface).
On the downside are the higher requirements for the local computer hardware (as compared
to the CGI approach again) which can result in slower execution speed.
The program supports all three request types of version 1.1.0 of the WMS standard.
Currently there is no built-in map projecting and datum conversion formulae. for hand-
ling different SRSs WMSClient relies entirely on the server to provide bounding boxes for all
Spatial Reference Systems in all layers in GetCapabilities response—when the server supplies
the bounding box co-ordinates in the units of a particular SRS then it is trivial to calculate
the co-ordinate parameters of the new request using only elementary arithmetics1 .
It also supports the composite (blended ) map images where the actual map, as shown
to the user, is composed of two or more raster images from multiple servers. This feature
mimics the behaviour of a cascading WMS server but purely on the client side—there is no
need for setting up an intermediate server, the computer user can select for blending any
servers he likes. When the user has selected the composite image then all following actions
(such as changing the scale) are carried out synchronously for all servers. The creation of
the composite image is accomplished by modifying the alpha channel (adding transparency)
of the upper images. The images must share a common co-ordinate system for blending.
The user interface of the application is not sophisticated but it offers complete control
over the process of selecting desired information layers, navigating, and making feature info
queries. Multi-level history is available: users can move back and forth in previously obtained
map images without reconnecting to the WMS server. On-line help is provided for quick
reference of the user. There is also menu entry for saving the map image in case there is
desire to process it further with other tools.
WMSClient is available for downloading at http://r-systems.ee:7007/WMSSvenska/
Applett/.
1
The first query is done using the bounding box co-ordinates, subsequent request are based on the user
interaction and the parameters of the previous request.
CHAPTER 5. WEB MAP SERVICE CLIENT PROGRAMS 49

Figure 5.2: WMSClient screenshot. Muuga harbour. The map image is blended from two
sources—Estonian Base Map from the WMS server of Estonian Land Board (background)
and a nautical chart from the WMS server of Estonian Maritime Administration (fore-
ground).
CHAPTER 5. WEB MAP SERVICE CLIENT PROGRAMS 50

5.3 WMS Client Toolkit


WMS Client Toolkit is an toolkit to facilitate the production of specialized WMS client
applications, developed by R-Süsteemid. The toolkit contains several components, written
in Javascript and PHP programming languages that can be used for making WMS client
applications, tailored for particular needs both in visual appearance and in functionality.
The applications would run in the web browser. The WMS client toolkit uses technology
that is only available in the Microsoft Internet Explorer (version 5.0 or higher) browsers,
thus restricting the potential users to people who can (and want to) use this particular breed
of browsers.
The PHP component makes HTML pages from the XML and Extensible Stylesheet
Language (XSL), using the Sablotron2 XML toolkit. XML data is fetched from the WMS
server (server capabilities XML or GML), necessary XSL files are provided with the toolkit.
Javascript is employed for user interface elements (dynamic HTML): menus, form pro-
cessing, etc. Javascript program is embedded into the HTML page and is executed in the
web browser (in the computer of the user).
The layout of the HTML page that forms the interface to the WMS client is not dictated
by the toolkit and is freely customizable by the creators of the client application using HTML
and Cascaded Style Sheets. One particular example is shown in Figure 5.3 on the next page.
All request types of WMS protocol version 1.1 are supported. The result of the Get-
Capabilites is stored inside the HTML page that represents the user interface, eliminating
the need of requesting capabilities XML from the WMS server for every action. The HTML
page, displayed to the user, usually includes the map image and several other user interface
elements for setting the query parameters, such as selecting the map layers from a list and
choose appropriate presentation style for each layer.
The toolkit provides also additional features for controlling the behaviour of the client
program:

• The GML output of a Web Map Service server is translated into HTML and shaped
into tabular form for easy perception by humans, the result can be seen in Figure 5.4 on
page 52.

• History buffer for storing recently loaded images. The images from the buffer are
displayed immediately, without querying the WMS server. User can move back and
forward in the history buffer.

• Combination of map images from multiple servers. The blending is done by altering
the alpha channel information, thus providing partial transparency of images (except
the bottom-most).

• Various navigation controls for panning and zooming the image, ‘drag-zoombox’ (drag
a box on the map and enlarge part of the map inside that box), requesting feature
information, setting and returning to anchor point. Text fields and drop-down boxes
are provided for directly changing the other parameters of the GetMap request (e.g.
co-ordinate system and the size of the image).

• Metainformation about the server and layers, extracted from the capabilities XML and
provided to the user in a nicely formatted table.

This toolkit cannot perform conversions between co-ordinate systems, for calculating
the co-ordinates for the GetMap and GetFeatureInfo requests it uses the same method as
2
http://www.gingerall.com/charlie/ga/xml/p sab.xml
CHAPTER 5. WEB MAP SERVICE CLIENT PROGRAMS 51

Figure 5.3: An example of a WMS client based on the WMS Client Toolkit—the map browser
of the registry of assets of RMK.

WMSClient: map co-ordinates are calculated from the <BoundingBox> element of the WMS
server capabilities. Therefore the clients that are based on this toolkit are usable only with
those WMS servers that set the <BoundingBox> element correctly for each map layer (The
WMS standards does not require the inclusion of this element into the description of a map
layer.).
Currently there are two applications built with the WMS client toolkit: EMA public map
service client program (Section 6.1) and map browser for the registry of assets of Riigimetsa
Majandamise Keskus /State Forest Management Agency/ (RMK) (Section 6.3). The EMA
CHAPTER 5. WEB MAP SERVICE CLIENT PROGRAMS 52

Figure 5.4: An example of the view, displaying the results of the GetFeatureInfo request.
The figure shows information about two map features, a buoy and a topmark, from the
server of the EMA.

public map service client program is located in the EMA server and is accessible at the URL
http://195.50.203.246/wmsc/. The map browsing client of the registry of assets of RMK
is located at http://r-systems.ee:7007/rmk/WMS/rapla.html. The user interface of this
application is shown in Figure 5.3.

5.4 FRED Map Editor


FRED is a map editor for X11 environment in Unix-like operating systems, developed in
R-Süsteemid by the author of this paper. FRED is built on top of the Frege map handling
library.
The map editor has clean graphical user interface which allows precise positioning and
zooming, simple selection of objects by pointing to the map or selecting from list of all
objects and quick display of important information about selected map objects. FRED also
sports an interactive, hypertext based help system.
The map editor uses Frege data model, which is structured closely after S-57 (see sec-
tion 3.1.3 on page 18). Supported editing functions include the creation, modification and
deletion of feature objects (object’s identifier, class name, attributes, relationships with
spatial objects), spatial objects (position and attributes, relationships with other spatial
objects). Topology is supported during all spatial editing operations.
FRED can simultaneously display unlimited number of vector maps. The order of maps
is flexibly adjustable by the user. The maps can be of different types (based on different
CHAPTER 5. WEB MAP SERVICE CLIENT PROGRAMS 53

Figure 5.5: Screenshot of the FRED map editor. Lasnamäe district, Tallinn. Street map
(line features) is shown on top of raster image (Estonian Base Map, area features selected),
fetched from the WMS server of ELB.

data dictionaries)3 , user can choose the visible object classes for each map separately. The
appropriate presentation library can be selected at map loading, presentation parameters
can be changed later for each map layer. One loaded map is always marked as active.
User can interact with the active map—select objects either from the list of all objects or
directly from the map (rectangular search), or perform other operations. One map can be
set editable, meaning that the user can modify the contents of that dataset. When no map
is marked editable FRED acts as a map viewer.
3
All maps are stored in Frege datafiles, the process of conversion from and to different file formats is
carried out by using external converter programs.
CHAPTER 5. WEB MAP SERVICE CLIENT PROGRAMS 54

FRED has the ability to load a raster image, acquired from a WMS server, onto the
‘bottom’ of all vector maps (i.e. vector maps are drawn on top of the image). The image is
‘glued’ to vector layers and moves automatically when user changes the viewing position or
scale (automatic GetMap request is sent to the selected WMS server with new parameters).
This feature extends the capabilities of the FRED map editor, giving the users of FRED an
additional layer of visible background information for positioning or digitizing purposes.
At the moment FRED supports only GetCapabilities and GetMap queries for the WMS
specification version 1.1.0. GetFeatureInfo request is currently not performed but is definitely
planned for future releases, as well as interaction with WFS servers (editing of remote
datasets).
A configuration dialog is provided for selecting WMS servers and changing query para-
meters (choosing layers and SRS for map queries). User can also choose to temporarily
switch off the display of the WMS image, e.g. for closer examination of data on a vector
layer.
The screenshot of FRED can be seen in Figure 5.5 on the preceding page.

5.5 Conclusion
This chapter drew some examples of client applications for Open GIS Consortium WMS
protocol. Today there are many programs available with different concepts for carrying out
the same operations. Everyone can choose the right tool according the purpose, like, and
technical restrictions. The browser-based clients are great for public services—no additional
software is required (the existence of web browser in every computer is a norm nowadays)
and their user interface often looks familiar (when the application is using only standard-
looking controls, like CubeVIEW). Other, more specialized WMS clients, like FRED, are
mostly intended for specialists, who want to integrate the output of a Web Map Service
server with other data in a professional environment. The downside of such tools is the
increased complexity, resulting in higher learning curve.
The user interface of HTML-based clients (like CubeVIEW and the clients that use WMS
Client Toolkit) is restricted to small set of elements, defined in the HTML standard. While
using other programming environments (Java, Microsoft Windows, X11) give more widgets
and amusements to the hands of an application developer it also restricts the potential clients
to those who can and want to use such programs on their favourite platforms.
The basic functionality is the same for all programs studied here, they all can request the
capabilities XML of the Web Map Service server, present the information to the user, and
display the map images, requested from the server with user-specified parameters. All clients
except FRED support also the GetFeatureInfo interface for providing more information
about the map content. WMSClient and WMS Client Toolkit do not contain program code
for co-ordinate transformations and rely on the server for providing accurate information
about the map’s co-ordinates.
I will reiterate the strong points of the covered applications:

CubeVIEW has support for highly structured data sources (WMS servers), it also provides
an indication of the area, visualized in the main image, on the world map which is
useful when one works with mid- and small-scaled maps. CubeVIEW support can
handle data in many co-ordinate systems.

WMSClient is a Java program that can run inside web browser (if it supports Java) or as
stand-alone program on various platforms. The most innovative solution is the possib-
ility to blend the images from multiple WMS servers into one map that is presented to
CHAPTER 5. WEB MAP SERVICE CLIENT PROGRAMS 55

the user. Such operation has similar effects as the server cascades but it is implemen-
ted completely on client side, thus eliminating the need of setting up an WMS server
(or troubling the server owners). It has also an image cache.

WMS Client Toolkit provides the blending of images from multiple servers, supports Get-
FeatureInfo request and keeps the cache of map images for fast reloading of previously
acquired images. It is a set of tools for creating applications with various functionality
and appearance, not a ready-made program.

FRED is a stand-alone program, of which the WMS client is only a tiny part, providing
the background information. The implementation details differ from other applications
covered here but the user interface and overall work-flow of the WMS part are similar
to CubeVIEW and other browser-based clients. It does not currently support GetFea-
tureInfo request nor the selection of styles for map layers but can perform co-ordinate
transformations between some co-ordinate systems.

The main conclusion drawn from this chapter is the great variance in the implementations
of WMS client programs, everyone should be able to choose the right tool for his necessities.
Chapter 6

WMS in Action

The focus of this chapter is on the projects that use the WMS standard for solving the
problems that are rising from the real necessities of the GIS users.
I will introduce three systems that are under development in the R-Süsteemid company.
Although there are many similar projects done around the world these undertakings are
among the first to use OGC specifications in Estonia. All these systems are currently (spring
2002) in prototyping or specification stage, hopefully they will be completed soon and become
the criteria for other similar projects.

6.1 Public Map Service of the Estonian Maritime


Administration
Estonian Maritime Administration (EMA) 1 is the sole producer of nautical charts in Estonia.
The purpose of this project is to provide an overview of the maps produced by the EMA
and also provide background maps for other services. The output of the map server is not
qualified for navigation on the sea.
There is a working prototype of the system which contains a couple of chart sheets,
available at http://195.50.203.246/wmsc/ (the WMS server is located at http://195.
50.203.246/F-Open-server/server.php). The final installment and testing of the public
map server is presently bogged due to the lack of funding from the EMA but should be
completed at the mid-2002.
The map data is currently provided in the L-EST co-ordinate system for compatibility
with other Estonian datasets (see next section), the fully functional server should offer the
same data also in unprojected (latitude/longitude co-ordinates) and perhaps also in Mercator
projection, traditionally used for nautical charts.
The system consists of a dedicated server computer with map database, web server, and
F-Open server WMS server software.
In the map database are stored all electronic nautical charts that are currently valid for
navigating in Estonian waters. Each chart is divided into several layers (navigational marks,
soundings, navigational restrictions, etc.). Some of the layers (e.g. navigational marks) would
be queryable, i.e. user can use the GetFeatureInfo interface of the WMS protocol for fetching
information about the objects on those layers (see Figure 5.4).
The map layers would be collected into high-level divisions, based on the scale and
purpose of the charts. The top layer displays the contours of the coastline of the Baltic Sea.
1
http://www.vta.ee/
CHAPTER 6. WMS IN ACTION 57

The structure of the layers can be seen in Figure 5.2 on page 49 (center part of the left
panel).
The system contains also the WMS client software which is based on the WMS Client
Toolkit, tailored to this particular project (direct connection to the Estonian Maritime Ad-
ministration’s WMS server, displaying of additional textual information about the charts,
etc.). The service itself (the cartographic data) is indeed perfectly usable with any WMS
client application.
The goals of this project are quite simple and could be achieved with other map servers
(with proprietary technology). The use of the open WMS protocol opens up additional
possibilities, for example the fictional information server of the yacht harbours of Estonia can
use these nautical charts as a background for their own map layers, showing the information
about the harbour facilities, hotels, or other information.

6.2 Data Exchange between Map Authorities


This section describes the pilot of the process of connecting the Geographical Information
Systems of two big map producing state agencies, EMA and Estonian Land Board (ELB) 2 .
EMA produces the nautical charts for Estonian waters, ELBis the producer of several large
spatial datasets (Land Cadastre, Estonian Base Map, Estonian Basic Map and others).
Although the products of the two organizations are principally different, they do share some
common objects, such as coastline, landmarks usable for navigation, etc.
Such situation mandates the checking and validating each other’s data, as the resulting
maps need to be consistent and not misleading. ELB and EMA are currently changing data
occasionally on compact discs. The aim of this project is to provide instant access to the
spatial databases of the other authority, eliminating the need of physically transporting the
data and providing on-line access to the most recent data.
The structure of the data, used by the agencies is quite different, let alone the tools that
are used for producing the maps. These factors make the interoperability between these
systems on the record (map feature) level very difficult and costly. Therefore the WMS
protocol was chosen as an intermediating agent, the WMS clients run in ubiquitous web
browsers, providing the raster image of the area of interest. When the cartographer has
layered images from two map servers on top of each other (either by using a cascading server
or by taking advantage of the composite image technology of the WMSClient or WMS Client
Toolkit-based application), he can immediately spot any differences between the datasets.
He can then make the GetFeatureInfo request for getting the GML representation of the
feature object from the distant server. The GML file is then analyzed and compared with
existing data. The process comes to the end with patching the map with new data that is
converted to suitable form.
The pilot was conducted during the end of 2001 and the beginning of 2002 for testing
the basic technology (setting up WMS servers, testing the connectivity, etc.).
ELB provided a test server, based on the MapServer. During the test period this server
was populated with data about the Estonian administrative units and a piece from the
Estonian Base Map3 . Estonian Maritime Administration used the public server prototype
(see previous section), based on the F-Open server software, containing some nautical charts.
The data in both servers were in the L-EST co-ordinate system. An example of the composite
map from these two map servers is shown in Figure 5.2 on page 49.
2
http://www.maaamet.ee/
3
For practical work the Estonian Basic Map will be used.
CHAPTER 6. WMS IN ACTION 58

The pilot was carried out on the public Internet without any access restrictions, the
production system would certainly need strict access restrictions for allowing only certified
personnel to access the databases.
The results of the pilot were positive but the completion of the whole project have yet
to happen. The reasons for delaying the project are more administrative than technical,
mostly dealing with the ownership of the data, the organization of process of modifications
in case differenced are spotted between a nautical chart and the Estonian Basic Map, and
other similar questions.
There exist additional problems, mostly related to data quality and different co-ordinate
systems being used, that need attention.
The L-EST system is used for the Estonian Basic Map and most other maps in Esto-
nia. Nautical charts are printed in Mercator projection with varying standard parallels for
different chart sheets, ENCs contain geographical co-ordinates. Co-ordinate system trans-
formation is therefore needed for spatial data, imported from outside (i.e. other sources than
the main database for given organization) which can cause some serious, hard-to-catch errors
if not carried out carefully. The problem grows larger when data more than two different
co-ordinate systems need to be integrated.
Every agency is responsible for the accuracy of the maps they produce. The requirements
for accuracy for nautical charts are precisely defined and stricter than for other maps. EMA
wants to be absolutely sure about the quality of the features they get from other sources, as
they would be legally liable for ship accidents, caused by errors on published charts.
A proposal for building a more complex system around the FRED map editor has been
made for the EMA by the R-Süsteemid company. This system would integrate into the
process of preparing nautical charts in the S-57 format, providing a raster image (obtained
from the WMS server) as a background to vector data in S-57 format. Using the WFS for
accessing the features may also prove practical in the future.

6.3 The Registry of Assets of RMK


R-Süsteemid created a web-based registry of assets for Riigimetsa Majandamise Keskus
/State Forest Management Agency/ (RMK) at the beginning of the year 2002. The applic-
ation used a MySQL4 database for storing the information and contained various forms for
processing and querying the data using the web browser. The database contained all assets
of the RMK—computers, vehicles, land parcels, leases, etc.
The registry includes several types of properties that contain spatial components: land
parcels, buildings, etc., leading to the idea of graphical representation of such information
as a map (originally the interaction with the registry was only via alphanumerical forms).
The choice of the WMS standard was quite natural, as the registry was already built for
WWW environment. The standard protocol would also provide beneficial in the future when
background maps can be fetched from other servers (e.g. Estonian Basic Map).
At present time a prototype of such system has been produced for demonstrating the
feasibility of this approach, containing the original registry program and database, a spatial
database, F-Open server WMS server and a special WMS client application, created with
the WMS Client Toolkit. The prototype includes data for only land parcels that are located
in Rapla county.
The original registry database contained insufficient information for visualizing land par-
cels, they were identified by only one co-ordinate pair and the cadastral identifier pointing
to the Land Cadastre, held by the ELB. The geometries of the land parcels were copied from
4
http://www.mysql.com/
CHAPTER 6. WMS IN ACTION 59

the Land Cadastre to a new PostGIS database5 , character (attribute) data originated from
the registry of assets of RMK. PostGIS database was chosen because it is free software (good
for minimizing the cost of a prototype system) and it supports the OGC Simple Features for
SQL ([42]) data model (see also Section 4.3.2). F-Open server loads data from spatial data-
base, attribute database (assets of RMK) and uses a specially created presentation library
for visualizing the data.
Currently two special operations have been added to the asset registry: WMS client is
opened with the map, showing the location of the land parcel when clicking on a button in
parcel’s (textual) information screen. The other innovation performs the opposite function—
clicking on the map can bring up the appropriate properties of the land parcel from MySQL
database. The latter function is based on the GetFeatureInfo request, the location of the
mouse pointer is transferred into co-ordinates, map server finds the object(s) and returns
the feature data, encoded in GML. The link into the registry program is done using the
identifier inside the feature data, returned by the WMS server.
The public demonstration of the application is available at http://r-systems.ee:7007/
rmk/WMS/rapla.html (screenshot is shown in Figure 5.3 on page 51). The map server
contains four layers: RMK land parcels and also the full land cadastre information for Rapla
county, forest compartments, and forest partitions for Vardi forest district (samples of real
data provided by RMK). The other three layers serve as additional background information
for giving context for the RMK land parcels. The map of administrative units from the test
server of ELB (see previous section) is used as the bottommost layer (not shown in Figure 5.3)
for a proof-of-concept for interoperability.

6.4 Conclusion
The projects described in this chapter indicate that the time is ready for using the Web
Map Service (WMS) protocol for solving real world problems. The OpenGIS Consortium
geospatial services standards can turn the web-based Geographical Information Systems into
a realty—you can use your maps everywhere, no matter where you are.
The public Internet map servers are not new players on the GIS arena and they are
certainly used not only for amusement purposes but also for real work. Using the WMS-
based public map servers would benefit all users, as they receive new qualities, such as the
exact georeferencing of the map images, extensive metadata, and most importantly the great
interoperability of the WMS servers. These enhancements build up from the ground set up
by earlier map servers, still preserving their strong points (cross-platform access via web
browsers, simplicity).
When the on-line system for spatial data exchange between Estonian Maritime Admin-
istration and Estonian Land Board is completed, it will certainly increase the quality of
maps, produced by those agencies. The system can be easily extended later to exchange
data between other map-makers as well.
The combination of the asset registry of the RMK with the WMS has been successfully
demonstrated. Besides the impacts to the registry project (increased efficiency in managing
the land parcels) it has shown the feasibility of this approach for other similar projects. The
task of adding spatial dimension to an existing database (turning an information system
into a GIS) is usually not trivial but the results are worth the trouble. The WMS protocol
can provide an easy and effective solution for web-based databases (also for others but more
work is needed for proprietary database interfaces) for the ‘spatialization’ problem.
5
Some sort of service will be used in the future for fetching cadastral data from the server of the Estonian
Land Board.
Chapter 7

Conclusions

It is widely recognized that standards are the key to address interoperability: spatial data
handling provides enough challenges and there is no need for creating artificial obstacles to
the process of using geospatial data. In this work I have advocated the importance of using
open standards for maximizing the interoperability in hierarchical and peer-to-peer systems.
Using open standards that are approved by international organizations standards is strongly
recommended for every GIS application.
It has been a pleasure to work in the front line of the GIS development. The current
thesis is one of the first publications (together with the previous technical report which
was co-authored by me ([29])) that covers the OpenGIS services in the context of present
Estonian reality.
The results of my research are summarized in the list below (more detailed discussions
of these results are given at the end of corresponding chapters):

• An overview of two basic methods of interoperability of Geographical Information


Systems, the exchange of files and the use of network based data services was given in
the third chapter. Introduction to the interoperability was provided together with the
descriptions of several widely used formats of spatial data files. Comparisons of the
described formats and suggestions for choosing a file format were also provided in this
chapter.
While there are many different formats and methods used for communication and
data exchange in GIS community, quite a large number of these do not qualify as
robust, adequate or do not have open specifications available. However, there exist
high quality open standards for both services and file formats. Open standards that are
created in co-operation of many participants should be preferred to proprietary, one-
company solutions, because internationally approved standards have adequate support
for metadata, their specifications are available for third-party implementations, and
they have usually more comprehensive and universal data models. From the file formats
covered in this thesis I suggest using the DIGEST standard, as it provides all previously
mentioned qualities (S-57 has very similar capacity but it is more specialized). GML
may prove adequate for applications that carry out only simple operations on data
(GML does not support topology in the current version) and make extensive use of
the XML technology.
As Geographical Information Systems are migrating from the desktop to a networked
environment the services will become more important forms of obtaining spatial data.
Therefore the focus was also on services in the chapter of interoperability. Most at-
tention was given to the family of web-based geospatial services provided by the Open
CHAPTER 7. CONCLUSIONS 61

GIS Consortium. Services are relatively new method for providing interoperability in
the GIS community but they will undoubtedly become more important in the future.

• Chapter 3 introduced two specifications of the OGC services—Web Map Service and
Web Feature Server. Chapters 4 and 5 moved from formal specifications to practical
realizations of the WMS standard, providing descriptions of several WMS server and
client applications, respectively. A more in-depth view of one WMS server program,
F-Open server, was also provided for better understanding of the WMS protocol and
software.

• The increased interoperability of Geographical Information Systems that resulted from


the adoption of an open standard (the OGC WMS—a service for providing georefer-
enced map images together with descriptive metadata in the WWW environment) was
demonstrated in this thesis.
Three examples of using the Web Map Service protocol for solving practical problems
were provided in Chapter 6. These projects demonstrated clearly that open standards
are very useful for building innovative software applications that can easily exchange
data with other systems.

I would also like to point out some trends and problems for future development. The work
on improving existing standards and standardization new areas should definitely continue.
Standard organizations should do even more co-operation for ensuring the compatibility of
different specifications.
Geospatial data services are getting more and more attention and are becoming im-
portant data sources these days. Hopefully there will be a network of WMS servers with
affiliations to other services (geocoders, gazetteers, geoparsers, etc.) for provision of miscel-
laneous geographical data. WFS and other OGC specifications will gain importance in their
respective segments, providing transparent access to omnipresent feature data stores. Some
of these services would be publicly accessible but most will probably be private servers with
restricted access for employees or paying customers (similar to the present situation with
data files).
For wider acceptance of WMS (and other standards) it is very important that the stand-
ard is accepted early by major data providers. In Estonia it would mean that e.g. the
Estonian Land Board should set up a WMS server for providing full coverage data for Es-
tonia. The map server should be either public or semi-public (access restricted for specific
organizations) and should contain a good large scale map (the Estonian Basic Map or Land
Cadastre, for example). The WMS protocol would guarantee the usability of such services
as a background information in many map servers that provide some map layers which are
informative but may look unnecessarily sketchy without additional background. The on-line
access to central WMS server would also mean that the latest version of database is used,
the owner of some smaller server can concentrate to the quality of his data without worrying
much about the timeliness of the background information.
Glossary

Abbreviated Terms
ANSI American National Standards Institute

CAD Computer Aided Design

CGI Common Gateway Interface

CSDGM Content Standard for Digital Geospatial Metadata

DBMS Database Management System

DCP Distributed Computing Platform

DIGEST Digital Geographic Information Exchange Standard

DGIWG Digital Geographic Information Working Group

ECDIS Electronic Chart Display and Information System

ELB Estonian Land Board

EMA Estonian Maritime Administration

ENC Electronic Navigational Chart

EPSG European Petroleum Survey Group

ESRI Environmental Systems Research Institute

FACC Feature Attribute Coding Catalogue

GIF Graphic Interchange Format

GIPSIE GIS Interoperability Project Stimulating the Industry in Europe

GIS Geographical Information System

GML Geography Markup Language

HTML Hypertext Markup Language

HTTP Hypertext Transfer Protocol

IEC International Electrotechnical Commission

IHO International Hydrographic Organization


GLOSSARY 63

IMO International Maritime Organization

ISFF Intergraph Standard File Format

ISO International Organization for Standardization

JPEG Joint Photographic Expert Group

ODBC Open Database Connectivity

OGC Open GIS Consortium

PHP PHP Hypertext Preprocessor

PNG Portable Network Graphics

RMK Riigimetsa Majandamise Keskus /State Forest Management Agency/

SDTS Spatial Data Transfer Standard

SGML Standard Generalized Markup Language

SLD Styled Layer Description

SRS Spatial Reference System

TIFF Tag Image File Format

TIGER Topologically Integrated Geographic Encoding And Referencing System

WBMP Wireless Bitmap

WCS Web Coverage Server

WFS Web Feature Server

WMS Web Map Service

WRS Web Registry Service

WWW World Wide Web

XML Extensible Markup Language

XSL Extensible Stylesheet Language

Glossary
Attribute A trait, quality or property describing a geographical feature.

Co-ordinate system A numerical method for representing locations on the Earth’s sur-
face.

Data dictionary The description description of the basic organization of a database. A


data dictionary typically contains a list of all files in the database, the names and types
of all records and metadata about the records and restrictions. Also known as feature
catalogue.
GLOSSARY 64

Dataset An identifiable collection of related data. One dataset (also data set) contains
specific data (and accompanying metadata) from a certain area.

Feature A set of points, lines or polygons in a spatial database that represents an abstrac-
tion of a real-world phenomenon (entity). The terms feature and object are often used
synonymously.

Feature catalogue See data dictionary.

GIS Information system dealing with information concerning phenomena associated with
location relative to the Earth.

Geomatics A technology and service sector focussing on the acquisition, storage, analysis,
dissemination and management of geographically referenced information for improved
decision-making.

Geocoder A system for assigning geographic co-ordinates to a street (postal) address.

Metadata Data describing and documenting data.

Object See feature.

Service A service is a capability which a service provider entity makes available to a service
user entity at the interface between the entities.

Topology The properties of spatial configuration which remain invariant under continuous
(in all directions) transformation. Also the branch of mathematics that investigates
those propeties.
Bibliography

[1] Alexander, L. Electronic Charts: What, How, and Why?, June 2000.
URL http://www.caris.com/S-57/electronic_charts.html 3.1.3

[2] Association for Geographic Information. The GIS Dictionary, 1999.


URL http://www.geo.ed.ac.uk/agidict/welcome.html 2.1, 2.1

[3] Aunap, R. Eestis kasutatavad koordinaatsüsteemid. Coordinate Systems used in Es-


tonia, April 2001.
URL http://www.geo.ut.ee/~raivo/ESTCOORD.HTML 2.2

[4] Booth, C. J. (Editor) The New IEEE Standard Dictionary of Electrical and Electron-
ics Terms. IEEE Std 100-1992. The Institute of Electrical and Electronics Engineers,
Inc., fifth edition, 1993. ISBN 1-55937-240-0. 2

[5] Borland Software Corporation. dBASE .DBF File Structure, 7 1998.


URL http://community.borland.com/article/0,1410,15838,00.html 3.1.1

[6] Cargill, C. F. Why Are We Doing This? Computer, volume 34, 10 pp. 116–117,
October 2001. ISSN 0018-9162. 3

[7] Conforti, F. The new DGN file format is the foundation for revolutionary changes
to MicroStation. MicroStation Manager, June 2001. ISSN 1057-9567.
URL http://www.msmonline.com/jun01/cover.htm 3.1.2

[8] CubeWerx Spatial Warehouse Architecture.


URL http://www.cubewerx.com/main_en/system_architecture.html 4.2, 5.1

[9] The Digital Geographic Information Exchange Standard (DIGEST). Part 1. General
Description. Edition 2.1, Digital Geographic Information Working Group, September
2000.
URL http://www.digest.org/documents/Part1_GeneralDescription.zip 3.1.4

[10] Discussion Paper: OpenGIS Styled Layer Descriptor Draft Candidate Implementation
Specification. Version 0.7.0. OpenGIS
r
Implementation Specification OGC 01-028,
Open GIS Consortium Inc., February 2001.
URL http://www.opengis.org/techno/discussions/01-028.pdf 3.2.3

[11] Egenhofer, M. J. and Mark, D. M. Naive Geography. In A. U. Frank and W. Kuhn


(Editors) Spatial Information Theory, volume 988 of Lecture Notes in Computer Science,
pp. 1–15. Springer, Berlin (Germany), 1995.
URL http://www.geog.buffalo.edu/ncgia/i21/ng/NG51.html 2.1
BIBLIOGRAPHY 66

[12] ESRI Shapefile Technical Description. An ESRI White Paper, Environmental Systems
Research Institute, Inc., July 1998.
URL http://www.esri.com/library/whitepapers/pdfs/shapefile.pdf 3.1.1

[13] Federal Geographic Data Committee. Content Standard for Digital Geospatial
Metadata. FGDC-STD-001-1998, National Spatial Data Infrastructure, Washington
(DC, USA), June 1998.
URL http://www.fgdc.gov/standards/documents/standards/metadata/v2_0698.
pdf 2.3

[14] Filter Encoding Implementation Specification. Version 0.0.7. OpenGIS


r
Implementa-
tion Specification OGC 01-067, Open GIS Consortium Inc., October 2001.
URL http://www.opengis.org/techno/RFC13a.pdf 3.2.4

[15] Frank, A. U. and Mark, D. M. Language Issues for GIS. In Maguire et al. [33],
chapter 11, pp. 147–163.
URL http://www.wiley.co.uk/wileychi/gis/Volume1/BB1v1_ch11.pdf 2.1

[16] Fuller, S. and Cox, T. Anatomy of a Forward-Looking Open Standard. Computer,


volume 35, 1 pp. 140–141, January 2002. ISSN 0018-9162. 3

[17] Galdos Systems Inc. Why GML?


URL http://www.galdosinc.com/technology-whygml.html 3.1.5

[18] Gatrell, A. C. Concepts of Space and Geographical Data. In Maguire et al. [33],
chapter 9, pp. 119–134.
URL http://www.wiley.co.uk/wileychi/gis/Volume1/BB1v1_ch9.pdf 2.1

[19] Geographic Information—Terminology. ISO Committee Draft CD 19104, ISO/TC 211,


November 1999. 3, 3.2

[20] Geography Markup Language (GML) 2.1.1. OpenGIS r


Implementation Specification
OGC 02-009, Open GIS Consortium Inc., April 2002.
URL http://www.opengis.net/gml/02-009/GML2-11.html 3.1.5

[21] Halaris, G., Hadzilakos, T., Kavouras, M., Panopoulos, G., Paraschakis,
I., Sellis, T., Tsoulos, L., and Zervakis, M. Interoperability and Definition of a
National Standard. Issues of Interoperability, Database Integration and Definition of a
National Standard for the Exchange of Digital Geographic Data. 1998. GIS PlaNET
98 International Conference on Geographic Information, Lisbon (Portugal).
URL http://www.dbnet.ece.ntua.gr/~timos/geodb/epaged.pdf 3, 3.1.4

[22] Hecht, L. Stand up for Standards. GEOEurope, volume 9, 9 pp. 38–42, September
2000. ISSN 0926-3403.
URL http://www.geoplace.com/ge/2000/0900/0900stn.asp 3.2.1

[23] International Hydrograhic Organization. IHO ECDIS Presentation Library


User’s Manual. International Hydrographic Bureau, Monaco, 3.2 edition, May 2000.
URL http://www.iho.shom.fr/general/ecdis/download/pslb03_2.pdf 3.1.3

[24] International Hydrographic Organization. Catalogue of the IHO Pulications


2002–2003. International Hydrographic Bureau, Monaco.
URL http://iho.shom.fr/publicat/free/files/P402En.pdf 3.1.3
BIBLIOGRAPHY 67

[25] International Hydrographic Organization. IHO Transfer Standard for Digital


Hydrographic Data. Edition 3.0. International Hydrographic Bureau, Monaco, Novem-
ber 1996. 3.1.3

[26] Krechmer, K. The Need for Openness in Standards. Computer, volume 34, 6 pp.
100–101, June 2001. ISSN 0018-9162. 3

[27] Krusberg, P. Ruumiandmete töötluse võimalusi (Eesti baaskaardi näitel). Magis-


tritöö, Tallinna Tehnikaülikool, Tallinn (Estonia), November 1997. In Estonian. 2.1

[28] Kuus, H. Functions for Handling Digital Maps. Diploma Project, Tallinn Technical
University, Tallinn (Estonia), June 1998.
URL http://home.anet.ee/~hkuus/diploma/index.html 4.3.2

[29] Kuus, H., Krusberg, P., and Rebane, R. Geoinformaatika standardid. Ülevaade,
Riigi Infosüsteemide Osakond, March 2002. In Estonian.
URL http://ats.riik.ee/amphora/home/projektid/gis/Standardid/HTML/
index.html 3.2.2, 7

[30] Laurini, R. and Thompson, D. Fundamentals of Spatial Information Systems. Num-


ber 37 in The A.P.I.C. Series. Academic Press, London (Great Britain), San Diego
(USA), New York (USA), 1996. ISBN 0-12-438380-7. 2, 2.1, 2.3

[31] Limp, F. Don’t Hit Warp Speed with the Wrong Equipment. GEOEurope, volume 8,
12 pp. 1822–42, December 1999. ISSN 0926-3403.
URL http://www.geoplace.com/ge/1999/1299/1299wm.asp 3.2.1

[32] Maguire, D. J. An Overview and Definition of GIS. In Maguire et al. [33], chapter 1,
pp. 9–20.
URL http://www.wiley.co.uk/wileychi/gis/Volume1/BB1v1_ch1.pdf 2

[33] Maguire, D. J., Goodchild, M. F., and Rhind, D. W. (Editors) Geographical


Information Systems: Principles and Applications. Longman, London (Great Britain),
1991. ISBN 0-582-05661-6. 15, 18, 32, 34

[34] Maling, D. H. Coordinate Systems and Map Projections for GIS. In Maguire et al.
[33], chapter 10, pp. 135–146.
URL http://www.wiley.co.uk/wileychi/gis/Volume1/BB1v1_ch10.pdf 2.2

[35] MapServer Homepage.


URL http://mapserver.gis.umn.edu/ 4.1

[36] Mark, D. M., Chrisman, N., Frank, A. U., McHaffie, P. H., and Pickles, J.
The GIS History Project. In Proceedings of the UCGIS Summer Assembly. Bar Harbor
(Maine, USA), June 1997.
URL http://www.geog.buffalo.edu/ncgia/gishist/bar_harbor.html 1

[37] Meitner, M. J., Cavens, D., and Sheppard, S. R. Academic Metadata Standards:
Getting Compliance Without Enforcement. In ESRI User Conference Proceedings.
July 2001.
URL http://www.esri.com/library/userconf/proc01/professional/papers/
pap934/p934.htm 2.3

[38] MicroStation 95 Reference Guide, chapter 18.


URL http://gdal.velocet.ca/projects/dgn/ref18.pdf 3.1.2, 3.1.6
BIBLIOGRAPHY 68

[39] Moore, T. Coordinate Systems, Frames and Datums, 2000.


URL http://granby.nott.ac.uk/~iszwww/coord1.htm 2.2

[40] Newell, R. G. and Sancha, T. L. The Difference Between CAD and GIS. Small-
world Technical Paper No. 2, GE Smallworld.
URL http://emea.smallworld.co.uk/support/techpaper/tp2.html 3.1

[41] Open GIS Consortium Inc. Overview of OpenGIS Implementation Specifications,


August 2001.
URL http://www.opengis.org/info/gisworld/20010911.TS.SpecOver.htm 3.2.2

[42] OpenGIS
r
Simple Features Specification For SQL. Revision 1.1. OpenGIS Project
Document OGC 99-049, Open GIS Consortium Inc., May 1999.
URL http://www.opengis.org/techno/specs/99-049.pdf 4.3.2, 6.3

[43] Painho, M., Peixoto, M., Cabral, P., and Sena, R. WebGIS as a teaching tool.
In ESRI User Conference Proceedings. July 2001.
URL http://www.esri.com/library/userconf/proc01/professional/papers/
pap910/p910.htm 3.2

[44] Prothman, B. Meta data. IEEE Potentials, volume 19, 1 pp. 20–23, February/March
2000. ISSN 0278-6648. 2.3

[45] Robinson, A. H., Morrison, J. L., Muehrcke, P. C., Kimerling, A. J., and
Guptill, S. C. Elements of Cartography. John Wiley & Sons, New York (U.S.A),
Chichester (Great Britain), Brisbane (Australia), 1995. ISBN 0-471-55579-7. 2.1, 2.2

[46] Roy, A. Web Mapping Solutions.


URL http://www.gisdevelopment.net/technology/gis/techgi0037pf.htm 3.2.1

[47] Rüütel, M. Meremõõdistussüsteemi tarkvara loomine. Diplomitöö, Tallinna


Tehnikaülikool, Tallinn (Estonia), June 1997. In Estonian. 4.3.2

[48] Scheuermann, W. What is ECDIS and what can it do?, October 1996.
URL http://www.sevencs.com/ecdis/whatisecdis.htm 3.1.3

[49] Snellen, I. ICTs, Bureaucracies, and the Future of Democracy. Communications of


the ACM, volume 44, 1 pp. 45–48, January 2001. ISSN 0001-0782. 1

[50] Snyder, J. P. Map Projections—A Working Manual. Number 1395 in U. S. Geological


Survey Professional Paper. United States Government Printing Office, Washington (DC,
USA), 1987. ISBN 9998 605067. 2.2

[51] STASYS Limited. WGS 84—World Geodetic System 1984.


URL http://www.wgs84.com/ 2.2

[52] The OpenGISTM Abstract Specification Topic 0: Abstract Specification Overview. Ver-
sion 4. OpenGISTM Project Document Number 99-100r1, Open GIS Consortium Inc.,
June 1999.
URL http://www.opengis.org/techno/abstract/99-100r1.pdf 3.1.5

[53] The OpenGISTM Abstract Specification Topic 5: Features. Version 4. OpenGISTM


Project Document Number 99-105r2, Open GIS Consortium Inc., June 1999.
URL http://www.opengis.org/techno/abstract/99-105r2.pdf 3.2.4
BIBLIOGRAPHY 69

[54] Theobald, D. M. Understanding Topology and Shapefiles. ArcUser, volume 4, 2,


April-June 2001.
URL http://www.esri.com/news/arcuser/0401/topo.html 2.1, 3.1.1

[55] United States Census Bureau. TIGER r


Overview, 2001.
URL http://census.gov/geo/www/tiger/overview.html 3

[56] Warmerdam, F. DGNLib: a Microstation DGN (ISFF) Reader.


URL http://gdal.velocet.ca/projects/dgn/index.html 3.1.6

[57] Warmerdam, F. Microstation DGN Feature Representation.


URL http://gdal.velocet.ca/projects/dgn/representation.html 3.1.2

[58] Web Feature Server Specification. Version 0.0.14. OpenGIS


r
Implementation Specific-
ation OGC 01-065, Open GIS Consortium Inc., October 2001.
URL http://www.opengis.org/techno/RFC13.pdf 3.2.4

[59] Web Map Service Implementation Specification. Version 1.1.1. OpenGIS


r
Implement-
ation Specification OGC 01-068r3, Open GIS Consortium Inc., January 2002.
URL http://www.opengis.org/techno/specs/01-068r3.pdf 2, 3.2.3, 3.2.3, 4.3.3

[60] Winter, S. and Frank, A. U. Topology in Raster and Vector Representation.


GeoInformatica, volume 4, 1 pp. 35–65, March 2000. ISSN 1384-6175.
URL ftp://ftp.geoinfo.tuwien.ac.at/winter/winter00topology.pdf 2.1

[61] Winter, S., van der Vlugt, M., and Frank, A. U. Open GIS and Interoperability
in Europe. In Proceedings of GIS Ostrava ’99. Areal VSB-TU Ostrava, Ostrava (Czech
Republic), 1999.
URL ftp://ftp.geoinfo.tuwien.ac.at/winter/winter99open.pdf 2.4

[62] XML Schema. Part 1: Structures. W3C Recommendation, World Wide Web Consor-
tium, May 2001.
URL http://www.w3c.org/TR/xmlschema-1/ 3.1.5, 3.2.4

You might also like